• Cyberpunk 2077 Fans May Want to Keep an Eye on MindsEye
    gamerant.com
    Cyberpunk 2077 may have gotten a bad rap when it was first released in 2020 due to its shoddy technical state, but there really isn't another game just like it. Although there is no shortage of sci-fi video games out there, it cannot be said that there are a lot in the cyberpunk genre specifically. Indeed, the amount of AAA action-adventure games that take place in neon-drenched, Blade Runner-style settings are surprisingly limited. Night City, Cyberpunk 2077's dystopian sci-fi setting, is perhaps the best example of such a location in the world of video games. The fictional Californian city feels like a living, breathing place that has its own character.
    0 التعليقات ·0 المشاركات ·92 مشاهدة
  • 8 Ways Rainbow Six Siege 2 Can Evolve Into The Ultimate Tactical Shooter
    gamerant.com
    Ever since its release, Rainbow Six Siege has redefined the tactical shooter genre by emphasizing teamwork, destructible environments, and a deep understanding of intricate maps. However, it's time for Siege to catch up with the evolving landscape and transform into a modern tactical shooter.
    0 التعليقات ·0 المشاركات ·100 مشاهدة
  • 0 التعليقات ·0 المشاركات ·104 مشاهدة
  • AMD confirms big reveal for RX 9070 GPUs on February 28, on-sale date is early March so it looks like a head-to-head clash with Nvidias RTX 5070
    www.techradar.com
    AMD RX 9070 vs Nvidia RTX 5070 GPU showdown is apparently on, I just hope Team Red doesn't disappoint with pricing.
    0 التعليقات ·0 المشاركات ·98 مشاهدة
  • Which insurance should you choose? A quick guide on the coverage you need
    www.fastcompany.com
    As a subject for delightful conversation, personal insurance ranks somewhere between polyp removal and credit default swaps. Which means most of us dont know what we dont know.No one likes to dwell on what might go wrong in the futurewhich is part of the reason why we all tend to regard insurance professionals with a healthy level of skepticism. But protecting yourself and your money from the unexpected has to be part of getting your financial house in order. Otherwise, a single bad event could erase all your hard work.To figure out what kinds of insurance you might need, start with the following basic rules of the insurance industry.Social benefit and private profitThe goal of insurance is to share risk among a large pool of people. If everyone pays a small amountknown as the premiumto their insurance company, the insurer assumes the risk of any one individual suffering a large loss. At that point, the insurance company will pay out to make that individual whole after the loss.But insurance companies are not there just as a social benefit. These companies are in business to make a profit. This means insurers make it their business to understand what kinds of losses are most likely to happen. And if something is more likely to occur, the insurance company will charge higher premiums for it.This is why life insurance for a nonsmoking 20-something costs pennies compared to the giant chunk of change the same insurance costs for a pack-a-day 58-year-old with diabetes. Its possible the young adult might die in a freak stamp-collecting accident and perhaps the smoker might live to 109but the odds are that the 20-something has decades of life ahead and the 58-year-old does not.Since it is more likely that the insurance company will have to pay out for the smokers life insurance policy relatively soon, the premiums for that policy are higher. This is how the insurance company protects its profits while still offering the payout benefits.Mo money, mo likelyThe insurance industrys understanding of probable outcomes can help consumers identify which policies they need. Specifically, if a personal insurance policy is expensive, that usually means the insurer thinks its likely it will have to make a payoutand that can indicate that you might need that kind of coverage.This is not a one-to-one correlation, of course. Just because a policy is expensive doesnt mean you need it. And some types of insuranceidentity theft insurance and renters insurance, for exampleare extremely helpful to have and generally low-cost.But understanding why insurers charge high premium prices can help consumers figure out which types of policies they might need. The most expensive types of insurance include the following.DisabilityThis kind of personal insurance helps pay a portion of your salary until youre able to go back to work, which can help keep you financially stable. While you might assume that youre unlikely to suffer a disability, since the most strenuous thing you do is staple Mr. Lumbergs TPS reports, remember that about 1 out of every 4 current 20-year-olds will become disabled before reaching retirement age. Thats why its expensive to purchase disability insurancebut also why its important.Auto insuranceCar crashes are the leading cause of death in the United States, with a total about 120 people killed per day in car accidents. Getting behind the wheel is the riskiest activity most Americans engage in on a daily basis, which means the insurance to protect you from that risk is also expensive. (The good news is that you can lower your risk and your auto insurance costs by driving like your dad: hands on 10 and 2, brake gently, check your mirrors, and assume everyone on the road is trying to kill you.)Life insuranceEven though you wont notice if you die without life insurance, any dependents who rely on your income will struggle if you pass away. And that likelihood is 100%, since none of us are getting out of this thing alive. Life insurance is cheapest for young and healthy individualswho are the least likely to need it or buy itand the price goes up with age and health problems.Homeowners insuranceThis kind of insurance not only covers damage to your home and possessions because of a covered disaster, but also liability for injuries or property damage experienced by a visitor to your home. (This is why dog owners, even apart from the growing trend to take out healthcare policies on our furry friends, may pay a higher premium than pet-free homeowners, and certain dog breeds are not covered at allsince they are more likely to bite a stranger.) And even though homeowners insurance covers damage caused by certain disasters, not all types of hazards are covered. In particular, flooding is a common hazard that isnt covered.Flood insuranceNearly no insurers include flood damage in homeowners or renters insurance policies. Instead, you may have to purchase a policy through the National Flood Insurance Program, which is a partnership between the federal government, insurance companies, and local communities to provide affordable flood protection. This is because floods are so likely to happen in so many areas that the federal government had to help subsidize the cost of flood insurance.While these are not the only hazards you should protect yourself against, these are the ones the insurance industry (and its army of statistics nerds) think are most likely to occur. That means its a good idea to start with the types of losses you are most likely to face.Protect your moneymakerIn addition to looking at which hazards are most likely, its also helpful to think about what valuables you have that would be most difficult to replace. For most people, these valuables fall into the same categories as the most expensive types of insurance: your earning potential, your life, and your home represent your most valuable assets.But its important to insure whatever assets you have that would be financially devastating to lose. For example, opera singers have been known to insure their voices, since they cant earn their living if they cant sing. More commonly, small business owners and freelancers often purchase professional liability insurance, also known as errors and omissions insurance, to protect themselves from lawsuits.Thinking about insurance as protection against financial loss can help you pinpoint what kinds of personal insurance you need most.Dont fear the reaper (or the insurance rep)Theres a reason no one invites Ned Ryerson to dinner: talking about the kinds of doom-and-gloom that insurance professionals know intimately is a major bummer.But personal insurance is an important part of a healthy budget. You need insurance to protect you from the risk of a devastating financial loss, which it does by spreading the risk among a pool of individuals and asking the insurance company to assume the financial risk.Understanding how insurers price their policies can help you figure out which types of personal insurance are most important, since the industry charges higher premiums to protect against the most likely losses. Thats why disability insurance, auto insurance, life insurance, home insurance, and flood insurance are among the most expensive types of policies available. Insurers know they are likely to have to make payouts on these policies, so they price the premiums accordingly.Consumers can also figure out the right coverage by thinking about what assets it would be financially devastating to lose. For most people, that includes their income potential, life, and home, but depending on your circumstances, you may also want to protect other important assets that you rely on or would be unable to replace.
    0 التعليقات ·0 المشاركات ·86 مشاهدة
  • We give safety ratings to cars and restaurants. Why not do the same to protect our digital lives?
    www.fastcompany.com
    The American economy runs on what are known as heuristics, a diverse array of mental short-cuts that help consumers make a dizzying number of choices to navigate the wild complexity of everyday life. These shortcuts help us select the restaurants we may choose to patronize, the cars we drive, the food we purchase, and the schools we attend and to which we send our children. We rely on scoring systems, certifications, and ranking methodologies to consider what movies to see, what music to listen to, and whether to purchase fair-trade products. These shortcuts come in many forms, from the complex (like the tools used to rate bonds and other financial products) to the straightforward (like the letter grades that many municipalities generate to inform consumers whether a particular restaurant follows safe food-handling practices).Sometimes these systems are managed and operated by the government, like the National Highway Traffic and Safety Administrations system for grading automobiles and trucks for their performance in crash tests, but often by private entities, like Consumer Reports. Sometimes the ratings are purely peer-to-peer and aggregated, like the ubiquitous five-star rating systems for ride-hailing companies or delivery services.In the end, consumers rely on these systems every day to make decisions great and small, to help make sense of a complex world where we are too easily prone to information overload.One area that cries out for a methodology that would provide consumers with critical details about the products and services they are using is one that is largely devoid of these types of shortcuts: our online life.We search, scroll, bank, shop, talk, text, stream, post, like, stan, and even hook up in the digital world. And we enter sites, download apps, communicate over platforms, access our financial information, and provide intimate details about our health and welfare without the slightest clue about what the entities with which we share such information do with it. The truth is, most will use it for their own profit and often sell it to data brokers: the third-party entities that, in turn, pass it along to other companies that might then use and abuse it, selling us products, pushing content to us we may not want, and perhaps even getting us to engage in behavior we might otherwise avoid if we were truly educated consumers about the uses and abuses of our digital data. AI only amplifies that influence.But what if there was a way to use the power of heuristics to protect our digital privacy through simple shortcuts that could give consumers basic information about how different sites, apps, and platforms were exploiting the digital activities they harvest from us?At present, some American states and the European Union have created rules of the road for the sunny information superhighway, as it was once called so quaintly in the 1990s. Instead of an information superhighway where consumers can travel at will,free of harm or surveillance, when we enter the digital world today, a better metaphor is the Upside Down: the shadowy, parallel world from the hit TV series Stranger Things, where entities with access to our digital lives create replicants of us that follow us around, always just below the surface, waiting to do us harm.We are already living in a world where we get asked to accept a particular companys cookies policy or its terms of service. These relatively light touch disclosure regimes are the product of laws and regulations passed around the world. The Europeans General Data Protection Regulation (GDPR) has largely set the global standard because tech companies do not want to have to ascertain when a particular consumer is subject to those regulations or not. And it is the GDPR, and the European Union, that we have to thank for those ubiquitous pop-ups that ask us to accept the companys cookies policy.But those rules actually mask what is going on under the hood. Companies can comply with the disclosure requirements by giving consumers the option of accepting their practices or not, and burying those disclosures in user agreements that are unintelligible to the average user. As a result, current practices in the digital world require a far more robust regulatory response than that which the relatively weak disclosure regimes that presently exist currently offer.Consumers are also routinely presented with complex terms of service, which few will read to the end, and even a smaller number will completely understand. Indeed, rare is the consumer who ever actually reviews these policies prior to entering a site or download an app. If they did, they would likely find few privacy-protective policies, if any. Instead, more likely than not, a review of those policies would reveal that the company engages in cross-site tracking, sells consumers information, and forces such consumers to go to arbitration even for violations of those very terms of service policies, among other things.What legal protections do exist on the internet actually largely protect companies, and not consumers. Laws like Section 230 of the Communications Decency Act insulate many companies that engage in activities online from being sued for the content on their sites. Courts, too, following federal law, largely enforce the terms of service that require that disputes about a companys actions must be resolved, not through the courts, but through arbitration. All of this is a result of a powerful tech lobby that not only fights any meaningful regulation of their activities but also complains that any government intervention will stifle innovation and the economic benefits and convenience these companies generate.Enter the ZoneBut there is another way, one that does not require the heavy hand of government, that can still foster innovation and put the power in the hands of consumers to drive business behavior and not the other way around. A more robust regulatory regime for the digital world could draw on the power of grading systems to send a clear message to consumers about the risks that particular apps or sites may pose to our digital privacy. It would provide this information to consumers in an easy-to-understand format that does not require a deep dive into the bowels of a companys end-user agreement, or a certificate in legalese. Instead, whenever a consumer accessed a site, app, or platform, that service would communicate whether it is protective of the consumers privacy or not.While there are many ways that a company can protect, or violate, a consumers privacy, and engage in activity that makes it unaccountable to that consumer should it breach their privacy, a simple, easy-to-understand system would grade companies on how well they do in terms of protecting their customers privacy or routinely violate it. That information would be communicated through one letter, a grade, that the company would have to reveal prominently as any consumer accessed the service. The consumer would then know, immediately, whether this is an entity that looks out for consumer privacy and which tends to exploit it. But where would such grades come from?Some grading systems are opaque, with the ultimate grade issued by a government agency, like the restaurant letter grades in New York City. One can assume that an A grade means that the restaurant meets basic quality standards. And its hard to find a restaurant worth their salt that does not have that A grade. In fact, anything less is usually enough to ward off many customers.In a regime for the digital world, one could adopt a type of digital zoning modelled after land-use restrictions in IRL. In land-use zoning, certain uses are permitted and others are excluded in particular areas or zones. You generally dont have a power plant or waste treatment facility abutting single-family homes. Thats because of zoning. If an area is zoned for particular uses, individuals and businesses that wish to engage in those uses are free to do so within it. Developers, government regulators, commercial establishments and residents can easily find out what is permitted and what is not from a predetermined description of particular zones. Anyone can comply with those restrictions, or find themselves facing litigation, fines, an order to stop what they are doing, and perhaps even dismantle any illegal development that has occurred.Zoning in the digital world could work much the same way. Privacy-protective uses will be clustered in the best zone; lets call it Zone A. In that zone, companies would not track a consumers activities on their site, not even keep personally identifying information unless it was necessary for their own purposes, and certainly would not sell such information to third parties. They would agree to stiff punishments for violations of their consumers privacy and allow those disputes to be resolved in a court of law, instead of forcing individuals to go through business-friendly arbitration settings of those businesses choosing, as many companies choose to do today. Ultimately, a company agreeing to provide this suite of privacy-protective practices by operating within Zone A would be able to market to its customers that they are doing so by displaying an A prominently on their home page, their apps site on an app store, or whenever a consumer starts to enter that site from their smartphone.If a company failed to provide these sorts of privacy protections, it would not receive that grade. Instead, it could choose from a number of different zones that would offer a different suite of protections along a spectrum, from best to worst. When a company provides some privacy protective measures, that would justify it displaying a higher grade, even if not an A. The system would cluster an array of practicescovering search, sale of data, monitoring user behavior, etc.and grade companies on the extent to which they meet the more privacy-protective practices or are more likely to take advantage of their customers. Those companies that are least protective of their customers data would earn an F. All companies would have to display their grade prominently whenever a consumer engages with that companys site, service, app, or platform. Consumers would have an immediate read on whether the company is looking out for the customer or abusing their data for its own benefit.While disclosure-based regimes are sometimes themselves abused, by, for example, companies making it difficult to understand what their policies are, or burying the important disclosure in legalese, a disclosure regime that is clear and easy to understand will put the power back in the hands of the consumer. Such a regime could create a race to the top, with companies vying to be more protective of their consumers data because they have to be completely transparent about their data privacy practices.Instead of stifling innovation and competition, digital zoning could actually encourage both, prompting companies to find ways to deliver their products and services in ways that are more protective of their customers interests and not less. Moreover, companies have a clear choice within this regime: no particular grade would be mandated. Companies would be free to do as they please with their customers dataprovided they are open and honest about their practices.What are the exact contours of this system and who would get to begin to cluster the different practices that determine the grade companies would receive? All of us. Legislators, technology companies, online safety and security experts, and consumers could engage in a dialogue around these issues to start to chart a course forward when it comes to our digital life that will encourage innovation that is protective of our privacy and does not simply see privacy as, at best, something to get around, or, worse, something to exploit.This type of robust and meaningful disclosure can occur without heavy-handed government intervention. Government will certainly have a hand in helping to write the rules of the road and setting the contours of the zones, with extensive input from a wide range of stakeholders, but it will not need to engage in extensive regulation of private companies. Of course, there will be a need to police company practices to make sure they are complying with the requirements of the letter grade they say they deserve, but that can be accomplished by stiff penalties, fines, and damages actions when companies misrepresent the types of protections they afford their customers. Such policing can come from state attorneys general and consumers themselves. It will also require strong whistleblower protections so that employees are free to come forward if the companies for which they work are not following the law, as well as stiff penalties for companies that engage in this sort of fraudulent behavior.Digital zoning would establish a clear and easy-to-understand approach to online privacy, empowering consumers while promoting corporate transparency and accountability. It could create a market-driven system that makes clear to consumers which companies protect their privacy and which might violate it. And it can enlist the government to police the boundaries of the zones, and not necessarily impose command-and-control policies from on high. Such a market-driven approach would place the consumers in the drivers seat and give them a clear sense of the rules of the roadand who is following them around.As technology becomes more and more present in our lives, its important we have a clearer way to know if the companies we do business with are harvesting our data or selling it to those who will use it for purposes we dont know, and would never accept if we knew it was happening. The time is right for us to better understand how technology serves us, rather than having such technology serve us up to anyone eager to exploit our data.Adapted fromThe Private Is Political: Identity and Democracy in the Age of Surveillance Capitalismby Ray Brescia. Published by NYU Press. Copyright 2025 byRay Brescia. All rights reserved.
    0 التعليقات ·0 المشاركات ·85 مشاهدة
  • 0 التعليقات ·0 المشاركات ·90 مشاهدة
  • Game Font Library reveals the typefaces of GTA, Halo and many more
    www.creativebloq.com
    This online resource is a treat for gamers and typography fans.
    0 التعليقات ·0 المشاركات ·87 مشاهدة
  • This New Algorithm for Sorting Books or Files Is Close to Perfection
    www.wired.com
    The library sorting problem is used across computer science for organizing far more than just books. A new solution is less than a page-width away from the theoretical ideal.
    0 التعليقات ·0 المشاركات ·98 مشاهدة
  • This AI Paper from Apple Introduces a Distillation Scaling Law: A Compute-Optimal Approach for Training Efficient Language Models
    www.marktechpost.com
    Language models have become increasingly expensive to train and deploy. This has led researchers to explore techniques such as model distillation, where a smaller student model is trained to replicate the performance of a larger teacher model. The idea is to enable efficient deployment without compromising performance. Understanding the principles behind distillation and how computational resources can be optimally allocated between student and teacher models is crucial to improving efficiency.The increasing size of machine learning models has resulted in high costs and sustainability challenges. Training these models requires substantial computational resources, and inference demands even more computation. The associated costs can surpass pretraining expenses, with inference volumes reaching billions of daily tokens. Moreover, large models present logistical challenges such as increased energy consumption and difficulty in deployment. The necessity to reduce inference costs without sacrificing model capabilities has motivated researchers to seek solutions that balance computational efficiency and effectiveness.Earlier approaches to address computational constraints in large model training include compute-optimal training and overtraining. Compute-optimal training determines the best-performing model size and dataset combination within a given compute budget. Overtraining extends training data usage beyond compute-optimal parameters, yielding compact, effective models. However, both techniques have trade-offs, such as increased training duration and diminishing performance improvements. While compression and pruning methods have been tested, they often lead to a decline in model effectiveness. Therefore, a more structured approach, such as distillation, is needed to enhance efficiency.Researchers from Apple and the University of Oxford introduce a distillation scaling law that predicts the performance of a distilled model based on compute budget distribution. This framework enables the strategic allocation of computational resources between teacher and student models, ensuring optimal efficiency. The research provides practical guidelines for compute-optimal distillation and highlights scenarios where distillation is preferable over supervised learning. The study establishes a clear relationship between training parameters, model size, and performance by analyzing large-scale distillation experiments.The proposed distillation scaling law defines how student performance depends on the teachers cross-entropy loss, dataset size, and model parameters. The research identifies a transition between two power-law behaviors, where a students ability to learn depends on the relative capabilities of the teacher. The study also addresses the capacity gap phenomenon, which suggests that stronger teachers sometimes produce weaker students. The analysis reveals that this gap is due to differences in learning capacity rather than model size alone. Researchers demonstrate that when compute is appropriately allocated, distillation can match or surpass traditional supervised learning methods in terms of efficiency.Empirical results validate the scaling laws effectiveness in optimizing model performance. The study conducted controlled experiments on student models ranging from 143 million to 12.6 billion parameters, trained using up to 512 billion tokens. Findings indicate that distillation is most beneficial when a teacher model exists and the compute or training tokens allocated to the student do not exceed a threshold dependent on model size. Supervised learning remains the more effective choice if a teacher needs to be trained. The results show that student models trained using compute-optimal distillation can achieve lower cross-entropy loss than those trained using supervised learning when compute is limited. Specifically, experiments demonstrate that student cross-entropy loss decreases as a function of teacher cross-entropy, following a predictable pattern that optimizes efficiency.The research on distillation scaling laws provides an analytical foundation for improving efficiency in model training. Establishing a methodology for compute allocation it offers valuable insights into reducing inference costs while preserving model performance. The findings contribute to the broader objective of making AI models more practical for real-world applications. By refining training and deployment strategies, this work enables the development of smaller yet powerful models that maintain high performance at a reduced computational cost.Check outthePaper.All credit for this research goes to the researchers of this project. Also,feel free to follow us onTwitterand dont forget to join our75k+ ML SubReddit. NikhilNikhil is an intern consultant at Marktechpost. He is pursuing an integrated dual degree in Materials at the Indian Institute of Technology, Kharagpur. Nikhil is an AI/ML enthusiast who is always researching applications in fields like biomaterials and biomedical science. With a strong background in Material Science, he is exploring new advancements and creating opportunities to contribute.Nikhilhttps://www.marktechpost.com/author/nikhil0980/This AI Paper from UC Berkeley Introduces a Data-Efficient Approach to Long Chain-of-Thought Reasoning for Large Language ModelsNikhilhttps://www.marktechpost.com/author/nikhil0980/Meta AI Introduces PARTNR: A Research Framework Supporting Seamless Human-Robot Collaboration in Multi-Agent TasksNikhilhttps://www.marktechpost.com/author/nikhil0980/This AI Paper Introduces CodeSteer: Symbolic-Augmented Language Models via Code/Text GuidanceNikhilhttps://www.marktechpost.com/author/nikhil0980/NuminaMath 1.5: Second Iteration of NuminaMath Advancing AI-Powered Mathematical Problem Solving with Enhanced Competition-Level Datasets, Verified Metadata, and Improved Reasoning Capabilities [Recommended] Join Our Telegram Channel
    0 التعليقات ·0 المشاركات ·63 مشاهدة