1 844.742.7448

Email Us

Remote Support

Contact

Posts tagged "Model"

  1. Home
  2. Posts tagged Model

Gemma, Google’s new open-source AI model, could make your next chatbot safer and more responsible
  • No Comments

Google has unveiled Gemma, an open-source AI model that will allow people to create their own artificial intelligence chatbots and tools based on the same technology behind Google Gemini (the suite of AI tools formerly known as Bard and Duet AI).

Gemma is a collection of open-source models curated from the same technology and research as Gemini, developed by the team at Google DeepMind. Alongside the new open-source model, Google has also put out a ‘Responsible Generative AI Toolkit’ to support developers looking to get to work and experiment with Gemini, according to an official blog post. 

The open-source model comes in two variations, Gemma 2B and Gemma 7B, which have both been pre-trained to filter out sensitive or personal information. Both versions of the model have also been tested with reinforcement learning from human feedback, to reduce the potential of any chatbots based on Gemma from spitting out harmful content quite significantly. 

 A step in the right direction 

While it may be tempting to think of Gemma as just another model that can spawn chatbots (you wouldn’t be entirely wrong), it’s interesting to see that the company seems to have genuinely developed Gemma to “[make] AI helpful for everyone” as stated in the announcement. It looks like Google’s approach with its latest model is to encourage more responsible use of artificial intelligence. 

Gemma’s release comes right after OpenAI unveiled the impressive video generator Sora, and while we may have to wait and see what developers can produce using Gemma, it’s comforting to see Google attempt to approach artificial intelligence with some level of responsibility. OpenAI has a track record of pumping features and products out and then cleaning up the mess and implementing safeguards later on (in the spirit of Mark Zuckerberg’s ‘Move fast and break things’ one-liner). 

One other interesting feature of Gemma is that it’s designed to be run on local hardware (a single CPU or GPU, although Google Cloud is still an option), meaning that something as simple as a laptop could be used to program the next hit AI personality. Given the increasing prevalence of neural processing units in upcoming laptops, it’ll soon be easier than ever for anyone to take a stab at building their own AI.

You might also like…

TechRadar – All the latest technology news

Read More

OpenAI’s new Sora text-to-video model can make shockingly realistic content
  • No Comments

OpenAI breaks new ground as the AI giant has revealed its first text-to-video model called Sora, capable of creating shockingly realistic content.

We’ve been wondering when the company was finally going to release its own video engine as so many of its rivals, from Stability AI to Google, have beaten them to the punch. Perhaps OpenAI wanted to get things just right before a proper launch. At this rate, the quality of its outputs could eclipse its contemporaries. According to the official page, Sora can generate “realistic and imaginative scenes” from a single text prompt; much like other text-to-video AI models. The difference with this engine is the technology behind it. 

Lifelike content

Open AI claims its artificial intelligence can understand how people and objects “exist in the physical world”. This gives Sora the ability to create scenes featuring multiple people, varying types of movement, facial expressions, textures, and objects with a high amount of detail. Generated videos lack the plastic look or the nightmarish forms seen in other AI content – for the most part, but more on that later.

Sora is also multimodular. Users will reportedly be able to upload a still image to serve as the basis of a video. The content inside the picture will become animated with a lot of attention paid to the small details. It can even take a pre-existing video “and extend it or fill in missing frames.” 

See more

You can find sample clips on OpenAI’s website and on X (the platform formerly known as Twitter). One of our favorites features a group of puppies playing in the snow. If you look closely, you can see their fur and the snow on their snouts have a strikingly lifelike quality to them. Another great clip shows a Victoria-crowned pigeon bobbing around like an actual bird.

A work in progress

As impressive as these two videos may be, Sora is not perfect. OpenAI admits its “model has weaknesses.” It can have a hard time simulating the physics of an object, confuse left from right, as well as misunderstand “instances of cause and effect.” You can have an AI character bite into a cookie, but the cookie lacks a bite mark.

It makes a lot of weird errors too. One of the funnier mishaps involves a group of archeologists unearthing a large piece of paper which then transforms into a chair before ending up as a crumpled piece of plastic. The AI also seems to have trouble with words. “Otter” is misspelled as “Oter” and “Land Rover” is now “Danover”.

See more

Moving forward, the company will be working with its “red teamers” who are a group of industry experts “to assess critical areas for harms or risks.” They want to make sure Sora doesn’t generate false information, hateful content, or have any bias. Additionally, OpenAI is going to implement a text classifier to reject prompts that violate their policy. These include inputs requesting sexual content, violent videos, and celebrity likenesses among other things.

No word on when Sora will officially launch. We reached out for info on the release. This story will be updated at a later time. In the meantime, check out TechRadar's list of the best AI video editors for 2024.

You might also like

TechRadar – All the latest technology news

Read More

Elon Musk says xAI is launching its first model and it could be a ChatGPT rival
  • No Comments

Elon Musk’s artificial intelligence startup company, xAI, will debut its first long-awaited AI model on Saturday, November 4.

The billionaire made the announcement on X (the platform formerly known as Twitter) stating the tech will be released to a “select group” of people. He even boasts that “in some important respects, it is the best that currently exists.”

It’s been a while since we’ve last heard anything from xAI. The startup hit the scene back in July, revealing it’s run by a team of former engineers from Microsoft, Google, and even OpenAI. Shortly after the debut on July 14, Musk held a 90-minute-long Twitter Spaces chat where he talked about his vision for the company. During the chat, Musk stated his startup will seek to create “a good AGI with the overarching purpose of just trying to understand the universe”. He wants it to run contrary to what he believes is problematic tech from the likes of Microsoft and Google. 

Yet another chatbot

AGI stands for artificial general intelligence, and it’s the concept of an AI having “intelligence” comparable to or beyond that of a normal human being. The problem is that it's more of an idea of what AI could be rather than a literal piece of technology. Even Wired in their coverage of AGIs states there’s “no concrete definition of the term”.

So does this mean xAI will reveal some kind of super-smart model that will help humanity as well as be able to hold conversations like a sci-fi movie? No, but that could be the lofty end goal for Elon Musk and his team. We believe all we’ll see on November 5 is a simple chatbot like ChatGPT. Let’s call it “ChatX” since the billionaire has an obsession with the letter “X”.  

Does “ChatX” even stand a chance against the likes of Google Bard or ChatGPT? The latter has been around for almost a year now and has seen multiple updates becoming more refined each time. Maybe xAI has solved the hallucination problem. That'll be great to see. Unfortunately, it's possible ChatX could just be another vehicle for Musk to spread his ideas/beliefs.

Analysis: A personal truth spinner

Musk has talked about wanting to have an alternative to ChatGPT that focuses on providing the “truth”, whatever that means. Musk has been a vocal critic of how fast companies have been developing their own generative AI models with seemingly reckless abandon. He even called for a six-month pause on AI training in March. Obviously, that didn’t happen as the technology advanced by leaps and bounds since then.

It's worth mentioning that Twitter, under Musk's management, has been known to comply with censorship requests by governments from around the world, so Musk's definition of truth seems dubious at best. Either way, we’ll know soon enough what the team's intentions are. Just don’t get your hopes up.

While we have you, be sure to check out TechRadar's list of the best AI writers for 2023.

You might also like

TechRadar – All the latest technology news

Read More

Meta takes aim at GPT-4 for it’s next AI model
  • No Comments

Meta is planning to meet, if not surpass, the powerful GPT-4 chatbots designed by OpenAI with its own sophisticated artificial intelligence bot. The company is planning on training the large language model (LLM) early next year, and likely hopes it will take the number one spot in the AI game. 

According to the Wall Street Journal, Meta has been buying up Nvidia H100 AI training chips and strengthening internal infrastructure to ensure that this time around, Meta won’t have to rely on Microsoft’s Azure cloud platform to train its new chatbot. 

The Verge notes that there’s already a group within the company that was put together earlier in the year to begin work building the model, with the apparent goal being to quickly create a tool that can closely emulate human expressions. 

Is this what we want? And do companies care?

Back in June, a leak suggested that a new Instagram feature would have chatbots integrated into the platform that could answer questions, give advice, and help users write messages. Interestingly, users would also be able to choose from “30 AI personalities and find which one [they] like best”. 

It seems like this leak might actually come to fruition if Meta is putting in this much time and effort to replicate human expressiveness. Of course, the company will probably look to Snapchat AI for a comprehensive look at what not to do when it comes to squeezing AI chatbots into its apps, hopefully skipping the part where Snapchat’s AI bot got bullied and gave users some pretty disturbing advice. 

Overall, the AI scramble carries on as big companies continue to climb to the summit of a mysterious, unexplored mountain. Meta makes a point of ensuring the potential new LLM will remain free for other companies to base their own AI tools on, a net positive in my books. We’ll just have to wait for next year to see what exactly is in store. 

TechRadar – All the latest technology news

Read More

Microsoft unveils Turing Bletchley v3: The AI model taking Bing to the next level
  • No Comments

Microsoft is working hard towards proving the 'intelligence' part in artificial intelligence, and has just revealed the latest version of its Turing Bletchley series of machine intelligence models, Turing Bletchley v3.

As explained in an official blog post, Turing Bletchley v3 is a multilingual vision-language foundation model, and will be integrated into many existing Microsoft products. If the name of this model sounds scary, don’t worry – let’s break it down. 

The ‘multilingual' part is self-explanatory – the model helps Microsoft products function better in a range of languages, currently standing at more than ninety. The ‘vision-language' part means that the model has image processing and language capabilities simultaneously, which is why this kind of model is known as ‘multimodal’. Finally, the ‘foundation model’ part refers to the conceptual and technical structure of the actual model. 

The first version of this multimodal model was launched in November 2021, and in 2022, Microsoft started testing the latest version – v3. Turing Bletchley v3 is pretty impressive because making a model that can “understand” one type of input (say, text or images) is already a big undertaking. This model combines both text and image processing to, in the case of Bing, improve search results. 

Incorporating neural networks 

The Turing Bletchley v3 model makes use of the concept of neural networks, which is a way of programming a machine that mimics a human brain. These neural networks allow it to make connections in the following manner, as described by Microsoft itself: 

“Given an image and a caption describing the image, some words in the caption are masked. A neural network is then trained to predict the hidden words conditioned on both the image and the text. The task can also be flipped to mask out pixels instead of words.”

The model is trained over and over in this way, not unlike how we learn. The model is also continuously monitored and improved by Microsoft developers. 

Where else the new model is being used

Bing Search isn’t the only product that’s been revamped with Turing Bletchley v3. It’s also being used for content moderation in Microsoft’s Xbox Live game service. The model helps the Xbox moderation team to identify inappropriate and harmful content uploaded by Xbox users to their profiles. 

Content moderation is a massive job scale-wise and often mentally exhausting, so any assistance that helps moderators actually have to see less upsetting content is a big win in my eyes. I can see Turing Bletchley v3 being deployed in content moderation for Bing Search in a similar manner.

This sounds like a significant improvement for Bing Search. The AI-aided heat is on, especially between Microsoft and Google. Recently, Microsoft brought Bing AI to Google Chrome, and now it’s coming for image search. I don’t see how Google doesn’t see this as direct competition in the most direct manner. Google still enjoys the greatest popularity both in terms of browser and search volume, but nothing is set in stone. Your move, Google. 

You might also like …

TechRadar – All the latest technology news

Read More

Apple may be considering a cheaper Vision Pro model – so people might be able to actually buy the headset
  • No Comments

The Apple Vision Pro is probably the most ambitious product to be announced this year, combining an interesting design and staggering price tag with some exciting-sounding tech that could make it really stand out from other virtual reality headsets (not that Apple wants the Vision Pro to be considered a VR headset). It features two 4K microLED panels that at the moment are only sourced from Sony, with the manufacturer capping annual display production at just a million units dedicated to the headset. 

This puts a spanner in the works, as not only will Apple be unable to produce as many units as possibly needed, but it also means the company has no negotiating power with component prices, as only Sony is making them. However, it seems like two Chinese suppliers are currently being evaluated to produce the microLED technology, which could enable mass production and hopefully, a cheaper model. 

According to The Information, two people “with direct knowledge of the matter” claim that Apple is “testing advanced displays” by two companies for possible inclusion in future models. 

A source cited in the article also hints at the possibility of a non-pro, more financially accessible, version of the Vision headset, stating that Apple is evaluating BOE’s and SeeYa’s – the two companies mentioned above –  displays for future models of both the Vision Pro and a cheaper headset internally code-named N109, which The Information previously reported was in an early stage of development.

The cheaper the better 

Apple already uses BOE for iPad and iPhone displays, so there is a good chance that they would collaborate again for Vison Pro panels. When the augmented reality headset was announced in June of this year, the steep price tag of $ 3,500 caused concern about who could actually afford to buy one.

In a time when people are concerned with the cost of living, who is this device actually for? During WWDC 2023 many people felt there was no clear audience for the Vision Pro, and at $ 3,500 not many people would be willing to shell out just to give the experimental technology a try. 

Hopefully, as Apple searches for cheaper display manufacturers and considers a more ‘basic’ Vision headset, it will give more people a chance to try out the impressive tech. Obviously, a cheaper alternative will have watered-down features, but I would rather spend half the price on a headset I can afford, that may be missing a few features than to be completely priced out of such exciting tech. 

You might also like

TechRadar – All the latest technology news

Read More

In an effort to surpass ChatGPT, Meta is open-sourcing its large language AI model
  • No Comments

In what could be an effort to catch up to – and maybe even surpass – the internet’s current AI darling ChatGPT, Meta will be open-sourcing its large language model, LLaMA 2, in collaboration with Microsoft.

This makes LLaMA 2 free to use for commercial and research purposes, putting it on the same level as OpenAI’s ChatGPT and Microsoft Bing.

The announcement was part of Microsoft’s Inspire event, where Microsoft revealed more about the AI tools making an appearance in the Microsoft 365 platform, as well as how much they would cost users. Meta’s explanation of the decision to make LLaMA 2 an open-source tool is to allow businesses, researchers, and developers access to more AI tools. 

Free roam LLaMA 

Rather than hiding the LLM behind a paywall, Meta will allow anyone and everyone to have access to LLaMA 2, thus opening up the door to more potential AI tools being built upon the model – which could be used by the general public.

This move also gives Meta a sense of transparency, something that has been conspicuously missing from some AI development projects; Meta notably also showed surprising restraint with regard to the release (or lack thereof) of its powerful new speech-generation AI, Voicebox.

According to The Verge, Meta will be offering the open-source LLM via Microsoft's Windows and Azure platforms, as well as AWS and Hugging Face. In a statement from the tech giant, the company states that “We believe an open approach is the right one for the development of today’s AI models, especially those in the generative space where the technology is rapidly advancing”. 

Meta seems to be pushing for a community angle with the move, hoping that interested user bases will help stress test and troubleshoot issues with LLaMA 2 quickly and collaboratively. It’s a bold move, and could be the perfect way to counteract Meta’s recent struggles in the AI arena. I can’t wait to see what people make with access to an open-source AI model.

TechRadar – All the latest technology news

Read More

Nvidia’s latest AI model could revolutionize games development
  • No Comments

AI could be about to make 3D modeling in a whole lot quicker and easier, if Nvidia has its way. 

Fresh from the company’s research lab, Neuralangelo is an AI model that transforms 2D video clips into “extremely high fidelity” 3D structures. 

“Creative professionals can then import these 3D objects into design applications, editing them further for use in art, video game development, robotics, and industrial digital twins,” the company explained in its latest paper.

Real-world high-fidelity 

Using videos as reference points is hardly anything new – even when you do throw relatively new and experimental AI into the mix. But, as Nvidia notes in its latest paper, “current methods struggle to recover detailed structures of real-world scenes.” In other words, they can replicate objects and scenes, but it won't feel lifelike. 

Read more

> This video maker's new AI editing tool picks your best takes for you

> These are the best laptop for AutoCAD deals

> Everyone is chucking AI in office software, and I'm not really sure why

And it’s that high fidelity Neuralangelo is promising to deliver. According to Nvidia Research, the minds behind the model, it's able to vividly recreate the textures of different materials, like glass, marble, and even roof shingling. The model has also proved capable of reconstructing building interiors and exteriors.

“The 3D reconstruction capabilities Neuralangelo offers will be a huge benefit to creators, helping them recreate the real world in the digital world,” said Ming-Yu Liu, senior director of research and one of the paper's co-authors.

Neuralangelo’s applications in art and games development are already clear: think of the vast cityscapes of Grand Theft Auto, the real-world historical setting of Assassin’s Creed. Whole virtual towns could soon be realized based on videos recorded on a smartphone. But it’s only a matter of time before AI models like these also form a core part of all the best architecture software and best landscape design software, if it really can streamline real-life reconstructions.

In a demo released by the company, results were admittedly impressive: highly detailed AI renderings of Michelangelo's David based on video footage that lived up to boasts of “detailed large-scale scene reconstruction from RGB video.” But we're still a way off from dropping AI 3D models straight into projects without requiring some finishing touches.

TechRadar – All the latest technology news

Read More

iPhone 12 release date could be staggered by model, with a mystery fourth iPhone too
  • No Comments

We're expecting the iPhone 12 announcement to take place in September, however there are claims the handset's release date might be as late as 2021, and a new iPhone 12 leak is giving us another possible release window.

Analyst Ming-Chi Kuo, who is perhaps the most reliable source regarding information on upcoming Apple products, has said (as reported by 9to5Mac) that while there are four iPhone 12 products coming later this year, they won't all be released at once.

Kuo seems to think that the two smaller iPhones – the 5.4-inch and 6.1-inch models – will enter mass production in September, with a larger 6.7-inch device only getting that far in October. 

It's not clear what the fourth iPhone will be, though it's possible it's also a 6.7-inch device (which we'll get to in a moment).

The diminutive (by today's standards) 4.7-inch iPhone SE (2020) – launched in mid-April – went into production in mid-March. This would suggest, going by Kuo's statement, that the new iPhone 12 handsets may not be available until October and November respectively. 

Unlike the iPhone 11 range, which became available all in one go, it seems the new iPhone 12 models may be released in waves.

This wouldn't be the first time Apple has staggered the release of its iPhone line. Wind the clock back to 2017, and the launch of the iPhone 8, 8 Plus and iPhone X, to see the later arriving a few months after the 8 and 8 Plus went on sale.

What are these four iPhones?

There's not been much word, from Kuo or otherwise, on what the four iPhones will be, but we've got a few ideas.

Based on precedent, we'd expect there to be a standard iPhone 12 model as well as an iPhone 12 Pro and iPhone 12 Pro Max. We've heard leaks about that latter phone, but that doesn't confirm the name will be used, because it was just provided by the leaker based on the iPhone 11 names.

The fourth device could be the anticipated 5G iPhone, which multiple rumors have stated is coming in 2020. We don't know much about the specs of this phone, or how it fits in the iPhone 12 range, but since Kuo only states three different sizes, it's possible the 5G iPhone will be the same size as the most top-end 4G device, perhaps 6.7 inches.

That would leave a curiously large gap between the phone sizes, with 0.7 inches between the small and medium devices and 0.6 inches between the medium and biggest, which is more of a gap than Apple has left in the past, but it's possible it's just trying to distinguish its devices more.

It's also possible this extra phone is the iPhone SE Plus, which we're been hearing about, and that would make sense with the 5.1-inch sizing. However that would leave the 5G iPhone mysteriously absent.

We'll find out more about the iPhone 12 range come September 2020, when we're expecting to see the handsets. Stay tuned to TechRadar before then for all the latest news, leaks and rumors.

TechRadar – All the latest technology news

Read More