Steel clashes and war drums thunder as a new age of battle dawns — one that will test even the mightiest Slayer. This GFN Thursday, DOOM: The Dark Ages — the bold medieval-inspired prequel to DOOM and DOOM Eternal — is available for GeForce NOW premium members, aka Ultimate and Performance members, to stream from the cloud at launch. Premium members can also slay in style with a free in-game reward. The stage is set and the crowd is buzzing — Capcom: Fighting Collection 2 is joining GeForce NOW at launch. Plus, get ready to take to the skies with Microsoft Flight Simulator 2024 coming to the cloud this week. And catch the latest GeForce NOW updates rolling out to members starting this week. The updates include quality-of-life improvements, following performance enhancements like 120 frames-per-second streaming for SHIELD TV to keep the cloud gaming experience at its best. It’s all part of another thrilling GFN Thursday, with five new games joining the cloud. Stand and Fight Keep your friends close and your enemies closer. DOOM: The Dark Ages is a dark fantasy and sci-fi single-player experience that delivers the searing combat and over-the-top visuals of the DOOM franchise, powered by the latest idTech engine. As the super weapon of gods and kings, shred enemies with devastating favorites like the Super Shotgun while wielding a variety of new bone-chewing weapons, including the versatile Shield Saw. Players will stand and fight on the demon-infested battlefields in the vicious, grounded combat the original DOOM is famous for. Take flight atop the new fierce Mecha Dragon, stand tall in a massive Atlan mech and beat demons to a pulp with the newly enhanced glory kill system. Only the Slayer has the power to wield these devastating tools of mayhem. Experience every gory detail, thunderous shield bash and demon-splitting kill in the cloud. No downloads, no waiting — just pure, uninterrupted DOOM action, wherever members want to play. SHIELD your eyes. GeForce NOW Ultimate or Performance members can now claim the DOOM Slayer Verdant skin reward, a fierce, ruthless-looking armor set that’s built for relentless slaughter. Those who’ve opted in to GeForce NOW’s Rewards program can check their email for instructions on how to redeem it. It’s available through Sunday, June 15, first come, first served. Step Into the Ring The fight continues. Capcom’s new fighting collection hits the stage — and the cloud. Choose from fan favorites like Capcom vs. SNK 2: Mark of the Millennium 2001 and Project Justice, as well as 3D action titles like Power Stone and Power Stone 2 in this collection of eight classic fighting games. Each can be played online or in co-op mode. Get back in the ring and duke it out in battles that everyone rumored but no one believed. Chase victory by streaming on GeForce NOW. Ultimate and Performance members enjoy higher resolutions and lower latency compared with free users for a true cloud-gaming edge. Game On Streaming from a powerful GeForce RTX gaming rig in the cloud enables GeForce NOW to deliver continuous improvements and new features that enhance members’ streaming experiences. This week, update 2.0.74 is rolling out, bringing several enhancements to the cloud. Members will see an upgraded library syncing feature for those using PC game subscription services like PC Game Pass and Ubisoft+, making it even easier to jump into games. Supported titles for these game services will now be automatically added to members’ “My Library” after resyncing their Ubisoft, Battle.net and Xbox connected accounts in the GeForce NOW app. This update follows the recent performance boost for SHIELD TV users in SHIELD Experience 9.2.1, now supporting up to 120 fps 1080p streaming for GeForce NOW Ultimate members. Those who prefer higher resolution over frame rates can continue streaming at up to 4K 60 fps. With such ongoing updates, GeForce NOW is making cloud gaming more seamless and accessible across devices. Fly Your Way Fly anywhere with the cloud. GeForce NOW brings a groundbreaking aviation experience to the cloud with Microsoft Flight Simulator 2024. Members can experience the game that redefines aviation simulation with unparalleled realism and global exploration. Pursue dynamic aviation careers through missions like Medevac, Search and Rescue, and Aerial Firefighting. Plus, compete in thrilling events such as the Red Bull Air Races. The game introduces advanced physics, enhanced aircraft systems and a groundbreaking flight planner for immersive gameplay. Explore an exceptionally detailed digital recreation of Earth, featuring handcrafted airports, landmarks, dynamic biomes, and real-time air and maritime traffic. With stunning visuals, diverse wildlife and realistic weather systems, Microsoft Flight Simulator 2024 offers unmatched experiences for pilots and adventurers. Ultimate and Performance members can play with GeForce RTX 4080-level performance with the highest frame rates and lowest latency. Ultimate members can elevate their adventures at up to 4K resolution and 120 fps for the most immersive rides in the sky. Fired Up for New Games It’s hammer time. Manage a medieval forge in Blacksmith Master, launching this week in the cloud. Find and hire the best staff and equip them with the right tools to optimize the business and train their skills over time. Design the shop for the best throughput, fulfill orders from across the kingdom to unlock new capabilities, and seek out new opportunities in the market as customers come looking for a variety of historically inspired items — from weapons and armor to tools and cooking utensils. Perfect the craft to become the Blacksmith Master. Look for the following games available to stream in the cloud this week: The Precinct (New release on Steam, May 13) Blacksmith Master (New release on Steam, May 15) Capcom Fighting Collection 2 (New release on Steam, May 15) DOOM: The Dark Ages (New release on Steam, Battle.net and Xbox, available on PC Game Pass, May 154) Microsoft Flight Simulator 2024 (Steam and Xbox, available on PC Game Pass) What are you planning to play this weekend? Let us know on X or in the comments below. Would you rather fight 1 demon the size of a castle or 100 normal-sized demons all at once? — NVIDIA GeForce NOW (@NVIDIAGFN) May 14, 2025

Author Colour prediction games have gained massive popularity across digital platforms in recent years, especially in regions like India. These games are easy to play, visually engaging, and promise quick results. But there's one question that every player or potential participant asks: Are colour prediction games based on luck or skill? This article dives deep into the mechanics, psychology, and patterns of these games to answer that question thoroughly. Understanding Colour Prediction GamesColour prediction games are a category of online games where players predict the outcome of the next color (often red, green, or violet) that will appear on a screen or wheel. If their prediction matches the result, they win and earn rewards. Many platforms and apps host such games, often labeling them as entertainment or lottery-style games.One such platform making waves is the 82 lottery game, which combines colour prediction with a digital lottery model. This has led to increased searches for 82 lottery download, as users are eager to try their luck or test their predictive skills on these platforms. The Appeal of Colour Prediction Games in IndiaIndia has witnessed a surge in casual gaming, and colour prediction games India has become a trending search term. With the rise of smartphone penetration and cheap internet access, millions are participating in these games, either for fun or for potential monetary gain.Games like lottery 82 not only offer a chance to win but also present an exciting challenge. The minimal entry barrier, instant results, and the possibility of winning make these games a top choice among young players and casual gamers. Are They Games of Luck or Skill?This is the core question — are these games purely chance-based, or do they require a level of skill? The answer isn’t entirely black and white. The Argument for LuckMany players argue that colour prediction games are fundamentally based on luck. Since the outcomes are typically generated by Random Number Generators (RNGs), each result is independent of the previous ones. This makes it almost impossible to accurately predict the next outcome using any pattern or strategy.Moreover, games like 82 lottery game or others on any colour prediction website often operate similarly to online lotteries, where luck plays the dominant role. In such cases, previous results offer little or no clue about the future outcome. The Argument for SkillOn the other hand, some seasoned players believe that there is a skill element involved, especially in understanding patterns, probability, and psychological tactics. These players use data analysis, game theory, and risk management strategies to make predictions that are slightly better than random guessing.Moreover, experienced users who frequently engage with platforms like lottery 82 often claim that consistent gameplay, proper bankroll management, and emotional control contribute significantly to success. Elements of Skill That May Influence ResultsWhile RNG systems are designed to be random, some subtle aspects of gameplay and human behavior can introduce a skill factor:Pattern Recognition: Some platforms may not use perfectly random systems, and users sometimes identify patterns.Probability and Statistics: Understanding odds and probability distribution helps make informed decisions.Risk Management: Smart players avoid all-in bets and stick to calculated risks.Experience: Like poker or fantasy sports, experience might provide an edge over new players.So while luck dominates, there is some room for skill—particularly on platforms where data transparency or algorithmic design allows for analytical prediction. Responsible Gaming Is KeyWhether you're playing the top colour prediction game or downloading the latest 82 lottery download, it’s crucial to play responsibly. These games can be entertaining and even rewarding, but they should not be seen as a reliable income source. Set a budget, know your limits, and avoid chasing losses.Several colour prediction websites now also offer responsible gaming features to help players manage their time and money effectively. This is a positive step, especially in markets like India, where regulatory frameworks are still evolving. Final Verdict: Luck or Skill?In conclusion, while colour prediction games may seem like they involve skill due to the decision-making element, the majority of their outcomes are driven by randomness. Luck plays a significant role, especially when RNG systems are involved. However, with proper strategies and experience, a small skill factor might influence outcomes in a player’s favor.For casual players, games like lottery 82 or the top colour prediction game provide thrilling entertainment. But always remember — treat them as games, not as a source of guaranteed earnings. Keywords Summary:82 lottery game: A popular colour prediction platform.82 lottery download: Refers to downloading the mobile version of the 82 lottery game.lottery 82: Another name or variant of the 82 lottery platform.top colour prediction game: Highlights the most popular platforms in this genre.colour prediction games India: Discusses the rise of such games specifically in the Indian market.colour prediction website: Refers to platforms offering these games online. Final ThoughtsThe rise of colour prediction games reflects our age-old fascination with prediction, chance, and challenge. Whether you're in it for fun or the thrill of winning, always remember the balance between luck and skill. Play smart, stay safe, and enjoy the colourful world of online predictions.

We know that Grand Theft Auto VI will have a massive impact on the video game industry when it launches, so it's no surprise that its delay into next year is also having a huge impact on the industry in 2025. Analytics firm Ampere Analysis has lowered its revenue and sales forecast for the video game console market, specifically for PlayStation and Xbox, The Game Business reports. Ampere's Piers Harding-Rolls expects PlayStation and Xbox to sell 700,000 fewer consoles than they would have if GTA VI were still being released this year. Harding-Rolls and Ampere also estimate that 21 million fewer game sales will be made, and that between hardware and software revenue, the console market is missing out on approximately $2.7 billion in revenue. Of course, the knock-on effect of all this is that expectations for 2026 are that much higher, but we're not yet halfway through 2025, and the year already looks drastically different from the one the video game industry thought it would have in January 2025. Without GTA VI in the holiday lineup, the one platform that will benefit, Harding-Rolls suggests, will be Nintendo. Harding-Rolls and Ampere Analysis also aren't alone in this suggestion, as DFC Intelligence reports that in its estimates, it sees Nintendo benefiting from Grand Theft Auto VI's delay, while PlayStation and Xbox have a much more muted year. Overall, Ampere now estimates that the industry will only grow by 0.9%, with a bigger 2.2% jump coming in 2026. Shortly after the news of Grand Theft Auto VI being delayed was announced, it was reported that the game had no chance of actually releasing in 2025. Unfortunate as that might be for fans hoping to play it this year, the delay doesn't seem to have really impacted excitement, as eager fans continue to salivate over the small bits of the game we have seen. Deal of the Day

Alibaba has unveiled Wan2.1-VACE, an open-source AI model designed to shake up how we create and edit videos.VACE isn’t appearing out of thin air; it’s part of Alibaba’s broader Wan2.1 family of video AI models. And they’re making a rather bold claim for it, stating it’s the “first open-source model in the industry to provide a unified solution for various video generation and editing tasks.”If Alibaba can succeed in shifting users away from having to juggle multiple, separate tools towards one streamlined hub—it could be a true game-changer.So, what can this thing actually do? Well, for starters, it can whip up videos using all sorts of prompts, including text commands, still pictures, and even snippets of other video clips.But it’s not just about making videos from scratch. The editing toolkit supports referencing images or specific frames to guide the AI, advanced video “repainting” (more on that in a sec), tweaking just selected bits of your existing video, and even stretching out the video. Alibaba reckons these features “enable the flexible combination of various tasks to enhance creativity.”Imagine you want to create a video with specific characters interacting, maybe based on some photos you have. VACE claims to be able to do that. Got a still image you wish was dynamic? Alibaba’s open-source AI model can add natural-looking movement to bring it to life. For those who love to fine-tune, there are those advanced “video repainting” functions I mentioned earlier. This includes things like transferring poses from one subject to another, having precise control over motion, adjusting depth perception, and even changing the colours.One feature that caught my eye is its ability to “supports adding, modification or deletion to selective specific areas of a video without affecting the surroundings.” That’s a massive plus for detailed edits – no more accidentally messing up the background when you’re just trying to tweak one small element. Plus, it can make your video canvas bigger and even fill in the new space with relevant content to make everything look richer and more expansive.You could take a flat photograph, turn it into a video, and tell the objects in it exactly how to move by drawing out a path. Need to swap out a character or an object with something else you provide as a reference? No problem. Animate those referenced characters? Done. Control their pose precisely? You got it.Alibaba even gives the example of its open-source AI model taking a tall, skinny vertical image and cleverly expanding it sideways into a widescreen video, automagically adding new bits and pieces by referencing other images or prompts. That’s pretty neat.Of course, VACE isn’t just magic. There’s some clever tech involved, designed to handle the often-messy reality of video editing. A key piece is something Alibaba calls the Video Condition Unit (VCU), which “supports unified processing of multimodal inputs such as text, images, video, and masks.”Then there’s what they term a “Context Adapter structure.” This clever bit of engineering “injects various task concepts using formalised representations of temporal and spatial dimensions.” Essentially, think of it as giving the AI a really good understanding of time and space within the video.With all this clever tech, Alibaba reckons VACE will be a hit in quite a few areas. Think quick social media clips, eye-catching ads and marketing content, heavy-duty post-production special effects for film and TV, and even for generating custom educational and training videos.Building AI models this powerful usually costs a fortune and needs massive computing power and tons of data. So, Alibaba making Wan2.1-VACE open source? That’s a big deal.“Open access helps lower the barrier for more businesses to leverage AI, enabling them to create high-quality visual content tailored to their needs, quickly and cost-effectively,” Alibaba explains.Basically, Alibaba is hoping to let more folks – especially smaller businesses and individual creators – get their hands on top-tier AI without breaking the bank. This democratisation of powerful tools is always a welcome sight.And they’re not just dropping one version. There’s a hefty 14-billion parameter model for those with serious horsepower, and a more nimble 1.3-billion parameter one for lighter setups. You can grab them for free right now on Hugging Face and GitHub, or via Alibaba Cloud’s own open-source community, ModelScope.(Image source: www.alibabagroup.com)Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.Explore other upcoming enterprise technology events and webinars powered by TechForge here.

These massive stargazing binoculars are hefty in both size and price, but is it worth splurging on the Extra Dispersion glass?

The big picture: One of the biggest challenges in virtual reality is mimicking the kind of sharpness we're used to seeing on TVs and monitors – but in VR, the screen is just inches from your eyes. While today's mainstream headsets like the Meta Quest 3 (2,064×2,208 per eye), PlayStation VR2 (2,000×2,040 per eye), and Apple Vision Pro (3,660×3,200 per eye) have made big strides, Samsung is now aiming to leap past those and push into truly lifelike territory with its latest next-gen displays. The company has crammed specs such as peak brightness levels of up to 20,000 nits and beyond-4K resolution into a new OLED display measuring no more than 1.4 inches across. Samsung is offering two versions of its headset. There's a 15,000-nit variant that delivers up to 5,000 PPI, and there's a 20,000-nit model boasting a 4,200 PPI. Both use a technology that Samsung likes to call OLEDoS (OLED on Silicon), where the OLED layers are deposited directly onto silicon wafers. That's what allows such dense pixel layouts and makes them ideal for compact VR form factors. As for why we'd want something that bright so close to our eyeballs, it's because VR headsets actually suffer massive brightness losses due to their optics. Pancake lenses, which help keep headsets slim and portable, are particularly inefficient, with around 90% of incoming light lost during transmission. Another factor affecting brightness is low-persistence illumination – a trick used to reduce motion blur. According to UploadVR, the combination ends up with your supposedly 5,000-nit display looking more like a 100-nit one. Samsung's new panels, with up to 20,000 nits of raw output, mean that even after all the optical losses, users could see 300 to 400 nits in the headset, which is still a meaningful jump compared to other options on the market today. The extra brightness could allow for not only sharper motion rendering but also more realistic daylight environments in mixed reality. Resolution-wise, one of Samsung's demo units boasts 3,888×3,888 pixels, while the other stretches beyond 4K per eye. Add in true RGB subpixel layouts (unlike the white OLEDs with color filters used in today's premium headsets), and you've got richer colors, deeper blacks, and better efficiency. Samsung also claims 99% DCI-P3 coverage, which is more than the Apple Vision Pro's already-impressive 92%. It all sounds almost too good to be true, but there is a catch. Manufacturing true RGB micro-OLED is significantly more complex and costly than the current standard. So for now, Samsung is positioning these panels as R&D and not yet ready for mass production. // Related Stories

New papal names often drip with meaning. Pope Francis, in 2013, named himself after Saint Francis of Assisi, signifying his dedication to poverty, humility, and peace. Pope Paul VI, in 1963, modeled himself after Paul the Apostle, becoming the first pope to make apostolic journeys to other continents. When Robert Francis Prevost announced on Saturday he would take the name Leo XIV, he gave an unexpected reason for his choice: the rise of AI. The most recent Pope Leo, Prevost explained, served during the Industrial Revolution at the end of the 19th century, and railed against the new machine-driven economic systems turning workers into mere commodities. Now, with AI ushering in a “new industrial revolution,” the “defense of human dignity, justice and labor” is required, he said. With his name choice and speech, Leo XIV firmly marks AI as a defining challenge facing our world today. But also embedded in the name is a potential path forward. Leo XIII, during his papacy, laid out a vision for protecting workers against tech-induced consolidation, including minimum wage laws and trade unions. His ideas soon gained influence and were implemented in government policies around the world. While it's still unclear what specific guidance Leo XIV may issue on artificial intelligence, history suggests the implications of his crusade could be profound. If he mobilizes the world's one billion Catholics against AI's alienating potential as decisively as his namesake confronted industrial exploitation, Silicon Valley may soon face an unexpected and formidable spiritual counterweight.“We have a tradition that views work from a theological perspective. It’s not simply burdensome; it’s where we develop ourselves,” says Joseph Capizzi, dean of theology and religious studies for The Catholic University of America. “Pope Leo XIV is going to be drawing on our tradition to try to make a case for finding work that dignifies human beings—even while making space for AI to do things that human beings will no longer be doing.” Rerum NovarumAt the heart of Leo XIV’s new name choice is Leo XIII’s formal letter Rerum Novarum, which he wrote in 1891. At the time, the Industrial Revolution was upending society. Mechanized production and factory systems generated unprecedented wealth and productivity, but led to the displacement of many agrarian jobs and people to move into overcrowded, unsanitary urban centers in search of work. The jobs there were grueling, unsafe, and paid terribly. The wealth gap widened dramatically, leading to massive social unrest and the rise of communist ideology. In the midst of these many challenges, Leo penned Rerum Novarum, an encyclical that marked the first major example of a pope commenting on social justice. In it, Leo wrote that “a small number of very rich men” had laid “upon the teeming masses of the laboring poor a yoke little better than that of slavery itself.” There now existed as “the gulf between vast wealth and sheer poverty,” he wrote. To combat this trend, Leo explored potential solutions. First, he rejected communism, arguing that workers had a right to the fruits of their own labors. But he also stressed the need for a living wage, time for workers for family and church, and the right to form Christian trade unions. “He was really championing the rights of workers,” says Dr. Richard Finn, director of the Las Casas Institute at Blackfriars, Oxford. In this colorized print from "La Ilustración Española y Americana," Pope Leo XIII directs a phonograph message to the American Catholic people on the occasion of his jubilee, in 1892. (Getty Images—LTL/Heritage Images)These ideas eventually caught hold. One of the first major advocates of minimum wage laws in the U.S. was the priest and economist John A. Ryan, who cited Pope Leo as a significant influence. Many ideas in his text “A Living Wage and Distributive Justice” were later incorporated into the New Deal, when Ryan was an influential supporter of President Franklin D. Roosevelt. In the 1960s, the Catholic Church eventually came out in support of César Chávez and the United Farmworkers (UFW), which Chávez told TIME in 1966 was the “single most important thing that has helped us.” In Australia, Rerum Novarum influenced political leaders who forged a basic wage in that country. And in Mexico, the Rerum Novarum spurred the creation of many Catholic labor unions and mutual aid societies. “It really shaped Catholic activism, with organizations working to ensure that Mexico was neither an unfettered capitalist country nor a Marxist state-owned state,” says Julia Young, a professor at the Catholic University of America. “It was successful in creating Catholic associations that were very politically vocal.” The Church and AIMore than a century after the industrial revolution, a similarly impactful technological revolution is unfolding, amidst many similar economic circumstances. “In terms of similarities between now and then, there was rural to urban immigration changing the workplace, widespread exploitation of workers, and seemingly growing poverty in urban areas,” Young says. “And so you had the church trying to respond to that and saying, ‘We have a different response than Marx or the robber barons.” While Leo XIV hasn’t yet explicitly called for any of the same measures as Leo XIII, it is clear that he believes the rise of AI necessitates some sort of counterweight. And his citing of Rerum Novarum also perhaps reveals a hunger to provoke widespread social change and offer a third path in a two-power arms race. “In our own day, the Church offers to everyone the treasury of her social teaching in response to another industrial revolution,” he said on Saturday. Across the world, people are expressing intense anxiety about AI causing job displacement. (Some economists contend that these fears are overblown, however.) Like in the industrial revolution, the initial spoils of AI are flowing to a few ultra-powerful companies. And AI companies have also reinforced some of the worst aspects of predatory global capitalism systems: OpenAI, for instance, outsourced some of its most grueling AI training to Kenyan laborers earning less than $2 an hour. Leo’s interest in this area continues that of Pope Francis, who became increasingly vocal about the threats to humanity posed by AI in his later years. Last summer at the G7 Summit, he called for an international treaty to regulate AI, arguing that it could exacerbate social tensions, reinforce dominant cultures, and undermine education. “We would condemn humanity to a future without hope if we took away people’s ability to make decisions about themselves and their lives, by dooming them to depend on the choices of machines,” he said. Some leaders have signaled the importance of prioritizing workers’ rights during the AI revolution, like Senator Josh Hawley. But until a coherent political movement emerges, moral leadership on human dignity in the face of AI may flow from the church, and Pope Leo’s outspoken leadership. “He’s saying AI is going to change the workplace—but it's got to change it in a way that fits with the dignity of employees,” says Dr. Finn. 

The data quality used in pretraining LLMs has become increasingly critical to their success. To build information-rich corpora, researchers have moved from heuristic filtering methods, such as rule-based noise removal and deduplication, to model-driven filtering, which leverages neural classifiers to identify high-quality samples. Despite its benefits, this approach still faces key issues: it lacks efficient validation mechanisms to assess data quality promptly and often relies on manually curated seed datasets that introduce subjectivity. While early datasets like C4 and Pile laid the groundwork for model development, recent efforts like RefinedWeb, Dolma, and DCLM have scaled significantly, incorporating up to trillions of tokens. Model-driven filtering has gained traction in these newer corpora for its ability to refine massive datasets and enhance LLM performance across downstream tasks. Nevertheless, the effectiveness of model-driven filtering is limited by the high costs and inefficiencies of current validation methods and the absence of clear standards for seed data selection. Recent datasets, such as FineWeb-edu and Ultra-FineWeb, have demonstrated improved model performance by using multiple classifiers to cross-verify data quality. These datasets outperform previous versions on benchmarks like MMLU, ARC, and C-Eval, indicating that refined filtering methods can enhance English and Chinese understanding. To further optimize this process, some studies propose using LLMs for multi-dimensional data evaluation via prompts or leveraging token-level perplexity scores. These innovations aim to lower computational overhead while improving data quality, ultimately enabling more effective training with fewer tokens.  Researchers from ModelBest Inc., Tsinghua University, and Soochow University developed an efficient data filtering pipeline to improve LLM training. They introduced a verification strategy that uses a nearly-trained LLM to evaluate new data by observing performance gains during final training steps, reducing computational costs. A lightweight fastText-based classifier further enhances filtering speed and accuracy. Applied to FineWeb and Chinese FineWeb datasets, this method produced the Ultra-FineWeb dataset, containing 1 trillion English and 120 billion Chinese tokens. LLMs trained on Ultra-FineWeb showed notable performance gains, confirming the pipeline’s effectiveness in improving data quality and training efficiency.  The study outlines an efficient, high-quality data filtering pipeline to reduce computational costs while maintaining data integrity. It begins by using a cost-effective verification strategy to select reliable seed samples from a candidate pool, which are then used to train a data classifier. Positive seeds are sourced from LLM annotations, curated datasets, textbooks, and synthesized content, while negatives come from diverse corpora. Classifier training avoids over-iteration, focusing instead on high-quality seed selection. A fastText-based classifier is used for scalable filtering, offering competitive performance at significantly lower inference costs compared to LLM-based methods, with preprocessing steps ensuring balanced, clean data input.  The models were trained using MegatronLM with the MiniCPM-1.2 B architecture on 100B tokens. Evaluations used Lighteval across English and Chinese benchmarks. The results show that models trained on Ultra-FineWeb consistently outperformed those trained on FineWeb and FineWeb-edu, individually and in mixed-language settings. Ultra-FineWeb-en achieved the highest English average score, while Ultra-FineWeb-zh improved performance on Chinese tasks. Ablation studies revealed that Ultra-FineWeb maintains balanced token lengths and benefits from efficient filtering strategies, highlighting its superior quality and effectiveness in improving model performance.  In conclusion, the study presents Ultra-FineWeb, a high-quality multilingual dataset comprising about 1 trillion English tokens and 120 billion Chinese tokens. Built upon FineWeb and Chinese FineWeb, it leverages a novel, efficient data filtering pipeline featuring a fastText-based lightweight classifier and a low-cost verification strategy. The pipeline enhances filtering accuracy, reduces reliance on manual seed data selection, and ensures robust performance with minimal computational overhead. Experimental results show that models trained on Ultra-FineWeb consistently outperform those trained on earlier datasets, demonstrating improved performance across benchmarks. The methodology ensures reproducibility and offers valuable insights for optimizing data quality in future LLM training.  Check out theTwitter and don’t forget to join our 90k+ ML SubReddit. Sana HassanSana Hassan, a consulting intern at Marktechpost and dual-degree student at IIT Madras, is passionate about applying technology and AI to address real-world challenges. With a keen interest in solving practical problems, he brings a fresh perspective to the intersection of AI and real-life solutions.Sana Hassanhttps://www.marktechpost.com/author/sana-hassan/Coding Agents See 75% Surge: SimilarWeb’s AI Usage Report Highlights the Sectors Winning and Losing in 2025’s Generative AI BoomSana Hassanhttps://www.marktechpost.com/author/sana-hassan/Rethinking Toxic Data in LLM Pretraining: A Co-Design Approach for Improved Steerability and DetoxificationSana Hassanhttps://www.marktechpost.com/author/sana-hassan/A Step-by-Step Guide on Building, Customizing, and Publishing an AI-Focused Blogging Website with Lovable.dev and Seamless GitHub IntegrationSana Hassanhttps://www.marktechpost.com/author/sana-hassan/NVIDIA AI Introduces Audio-SDS: A Unified Diffusion-Based Framework for Prompt-Guided Audio Synthesis and Source Separation without Specialized Datasets

Want to keep a scanner on demand? Since we don’t really have space for desktop scanners nowadays, SwiftScan VIP is the next best thing. A lifetime subscription to this handy scanning app is currently just $41.99 with code TAKE30 through June 1. Digitize your documents with just a few taps While the chunky scanners of yesteryear are hard to come by, the need to scan documents is still very much present. SwiftScan VIP serves as the 2025 version of that massive scanner, allowing you to create high-quality scans right from your smartphone. Whether you need to save a receipt, fill out a form, scan QR codes, or snap a barcode, SwiftScan can help. It automatically identifies the document edges and captures, then crops your pages, providing you with premium-quality scans that start at 200dpi. If you need to punch something up, there are color filters, auto-optimization, and blur reduction enhancement options. If you need to scan multiple pages, SwiftScan lets you save them as a single document with Multi-Page scans. Choose to save your file as a JPEG or PDF. If you pick PDF, you can also use SwiftScan VIP as a PDF editor—drawing, highlighting, or erasing text in your scans. It also makes it easy to add signatures to PDF forms, as well as reorder, add, or remove pages. Once you’re done scanning and editing, you can send files via email, fax, or directly to your cloud storage provider. Score your own lifetime subscription to SwiftScanVIP for just $41.99 with code TAKE30 through June 1. StackSocial prices subject to change.

Over the past few years, Okta has stated its commitment to ending the threat of identity-enabled cyber crime and attacks. As part of its Secure Identity Commitment, Okta has been keen on “elevating our industry’ by accelerating its capabilities and embracing new technology, such as AI, and und the digital transformation of nonprofits and advance inclusive pathways into tech. Therefore, when an announcement was made of a standard around identity security in security-as-a-service (SaaS) applications, it was worth taking notice. Named the Interoperability Profile for Secure Identity in the Enterprise (IPSIE), the concept is of an open standard which provides a framework for SaaS companies to enhance the end-to-end security of their products across every touchpoint of their technology stack. Announcing it in October 2024, Okta CEO and co-founder Todd McKinnon said there is a “need [for] massive standardisation” and “move to a world where every app, every device, every workload all speak a common language”. McKinnon said that by adopting IPSIE, users will get complete visibility into their identity environment and the threat surface, and they can provide access to the right applications at the right time and take real-time actions in response to threats. Okta’s announcement stated that the point of IPSIE is to “foster a more open, consistent, flexible SaaS ecosystem by empowering organisations to adhere to a higher level of security, more seamlessly and efficiently integrating among tech stacks”. This open standard will provide the framework for any enterprise application to be discoverable and governable. By adopting IPSIE, users will be able to gain complete visibility across the identity threat surface, enable consistent security outcomes across SaaS applications, and build secure-by-default SaaS applications more seamlessly and efficiently. On that final point, Okta states that any app built to the IPSIE standard adheres to a higher level of security by ensuring that it can be governed, have entitlements managed, can support multi-factor authentication and posture management, as well as feature real-time Universal Logout. So far, 50 enterprise SaaS applications have joined the cause and integrated with IPSIE – including Google, Microsoft Office 365, Slack and Salesforce – to support modern identity best practices aimed at enhancing security and reducing operational burden. Harish Peri, senior vice-president of product marketing at Okta, tells Computer Weekly that IPSIE is a way to ensure that every app and API conforms to a standard whereby its identity can be secure: “We are leading the way with the OpenID foundation, and we’re part of the working group for the creation of IPSIE interoperably profiled for secure identity of the enterprise.” Far from working alone, Okta has enlisted members of the OpenID Foundation to create the IPSIE Working Group, which will develop profiles of existing specifications with a primary goal of achieving interoperability between independent implementations. Gail Hodges, executive director of the OpenID Foundation, says that while the development of the IPSIE was initially getting off the ground in this first year, she felt the concept was “great”, adding: “I’m really encouraged as the foundation is moving more and more towards lining up specifications; like a lot of our work internally, they’re intended to kind of sync up with each other so that you could layer specifications on top of each other. “I see the work of IPSIE and a group of subject matter experts looking to do exactly that – line up the specifications together. So there’s even more consistency in how those specifications are configured, so there will be even greater benefits of interoperability and security associated with deploying a more complex stack. I think it’s fantastic.” Shiv Ramji, president of customer identity cloud at Okta, says the ultimate ambition with IPSIE is to “make it easy for customers to choose the right default path, which is to be secure, and I think they’ll do that if the value is clear to them, and, over time, it will be”. The concept of IPSIE from Okta is to gain industry-wide adoption, but Ramji was keen to make the point that Okta is “one participant”, and if every participant adopts the standards, “we will deliver better security outcomes for the entire software as a service ecosystem”. One factor Ramji stressed is the support for Universal Logout. Okta describes this as a concept where you can terminate users’ sessions, and their tokens, for supported apps when your identity threat protection identifies a risk change. Specifically, a user session is the time during which a user is authenticated and authorised to access apps secured by Okta, while an app session refers to sessions that an app generates to allow users to access the app’s resources. Universal Logout can be configured to terminate a users’ sessions in generic Security Assertion Markup Language (SAML) and OpenID Connect (OIDC) apps. Stephen McDermid, EMEA chief security officer at Okta, says the concept of Universal Logout will help to mitigate and minimise risks, “so that you’re not waiting for your SOC or your SIEM solution to respond in real time”. He adds: “I think the fact that there’s talk about the risks that IPSIE is trying to address reassures me that we’re going in the right direction for us – and for other vendors as well. The more vendors we can get to agree to it, the better the solution becomes.” This is why SaaS companies are integrating Okta’s software development kit, Ramji says, with companies now adopting this, “we’re changing the type of integrations that we do with these SaaS applications because we can do signal sharing”. In terms of integrations, Ramji says there were more than 150 in April 2025, and users “are asking us what are the ways they can support the adoption of these standards”. Out of those 150 integrations, is this something that the customer can implement on their own, rather than waiting for Salesforce, for example, to do it, for them? Ramji says if a user is using Auth0 today, they can switch IPSIE and Universal Login on and go into their Okta dashboard to enable the Universal Logout cable. “They have to enable it to opt in, as it’s an opt-in mechanism,” he says. “It’s easy to turn it on. As we roll this out initially, a lot of this will be opt-in, and then over time we can look at ways to make that easier, or maybe look at other options, but for now, it’ll be opt-in.We don’t want behaviours in companies where their applications where users are being logged out without working it out, so this is a deliberate thing that they need to roll out.” Peri says Okta’s largest existing customers asked, “How soon can you get all of our apps IPSIE-fied?”, and levels of IPSIE are being defined, but he adds that this is not an Okta-driven initiative or about asserting dominance, but “about doing the right thing for the industry, as the more people that are in it, the better is for everybody”. So, how well will IPSIE be adopted? Computer Weekly contacted a number of other authentication suppliers to find out. Chris Anderson, duo product CTO at Cisco, confirms that the firm had joined the IPSIE Working Group, which aims to develop profiles of existing specifications and achieve interoperability between independent implementations, stating: “While it’s still early days, we believe that interoperability across standards is key to greater success in identity security.” Andras Cser, vice-president and principal analyst at Forrester, says that standards that anyone can implement, proposed by one supplier, generally “do not fare very well”, but with the backing of the working group and OpenID Foundation, could work out well. He points at the example set by the FIDO Alliance, which “started out as a bunch of vendors coming together”. However, Cser believes that if IPSIE could follow FIDO’s lead, then it has a chance to work. “The use case behind FIDO was a lot smaller than IPSIE, it was just authentication and second factor and biometrics, that was the design and try not to boil the ocean,” he says. “Single sign on, logout and token verification are largely resolved by SAML and OpenID, and there’s a scanner for those things. “There’s also a very concrete and distinct use case behind sharing risk signals – there’s a new login from a new IP address, from a new device and that makes a lot of sense.” He claims that single sign on, token revocation and logout have been resolved, while user lifecycle management, and while other areas are being addressed further down the line. He adds that IPSIE is trying to resolve things, “30% of which are not solvable in the security domain only, 60% are addressed by other standards, and 10% is the key part of what IPSIE is trying to do”. Less than a year since its announcement, the conversations around IPSIE suggest it will take a long time to gain full traction and industry adoption, but there is persistent positivity on the side of Okta, its main supplier driver: the criticism comes from it being too broad and “putting everything in the kitchen sink”. Time will tell, but all revolutions need to start somewhere. Read more about identity management and SaaS applications Identity and access management tools and features for 2025. How to build an effective IAM architecture. Seven cloud IAM challenges and how to address them.