Low-rank adapters have become a standard approach for efficiently fine-tuning large language models (LLMs), but they often fall short of achieving the performance of full fine-tuning. We propose a method, LoRA Silver Bullet or LoRA-SB, that approximates full fine-tuning within low-rank subspaces using a carefully designed initialization strategy. We theoretically demonstrate that the architecture of LoRA-XS, which inserts a trainable (r x r) matrix between B and A while keeping other matrices fixed, provides the precise conditions needed for this approximation. We leverage its constrained update space to achieve optimal scaling for high-rank gradient updates while removing the need for hyperparameter tuning. We prove that our initialization offers an optimal low-rank approximation of the initial gradient and preserves update directions throughout training. Extensive experiments across mathematical reasoning, commonsense reasoning, and language understanding tasks demonstrate that our approach exceeds the performance of standard LoRA while using 27-90x fewer parameters, and comprehensively outperforms LoRA-XS. Our findings establish that it is possible to simulate full fine-tuning in low-rank subspaces, and achieve significant efficiency gains without sacrificing performance. Our code is publicly available at https://github.com/RaghavSinghal10/lora-sb.

The nation's preeminent organization dedicated to improving heart health has endorsed the lesser-known'portfolio diet'. Much like diversifying a stock portfolio with different promising investments, the'portfolio diet' involves incorporating various healthy dietary patterns together. It was invented by researchers from Harvard University T.H. Chan School of Public Health and is made up of a range of cholesterol-lowering foods. It's not as well-known as other popular diets such as the Mediterranean diet or the DASH diet, but it shares many similarities. It's not as well-known as other popular diets such as the Mediterranean diet or the DASH diet, but the'portfolio diet' shares many similarities For instance, followers are encouraged to swap in plant-based proteins instead of red meat and eat lots of complex fibrous foods like oatmeal and healthy fats like nuts.

In May, Sputnik International, a state-owned Russian media outlet, posted a series of tweets lambasting US foreign policy and attacking the Biden administration. Each prompted a curt but well-crafted rebuttal from an account called CounterCloud, sometimes including a link to a relevant news or opinion article. It generated similar responses to tweets by the Russian embassy and Chinese news outlets criticizing the US. Russian criticism of the US is far from unusual, but CounterCloud's material pushing back was: The tweets, the articles, and even the journalists and news sites were crafted entirely by artificial intelligence algorithms, according to the person behind the project, who goes by the name Nea Paw and says it is designed to highlight the danger of mass-produced AI disinformation. Paw did not post the CounterCloud tweets and articles publicly but provided them to WIRED and also produced a video outlining the project.

By now, you've heard of ChatGPT--or more likely, you've heard that it's coming to take your job whether you're a programmer, journalist, musician or almost anything else. The OpenAI chatbot has been accessible to all for mere months, yet it has already amassed millions of users, impressing with its ability to write everything from code to essays and lyrics. You may have also heard that ChatGPT is about to set a fire under an already bubbling-hot cyber threat landscape, helping scammers write engaging, convincing and grammatically correct phishing emails or perfect malware code in seconds. While ChatGPT is certainly impressive, do we really need to worry about it upping the ante on already menacing threats? In short, the answer is no--but there's little to celebrate about that.

Machine Learning's Large Language Models (LLMs) -- like ChatGPT, GPT3 and BERT -- have recently captured the attention of the world. Put simply, LLMs are artificial intelligence (AI) tools that read, summarize, translate and generate text. They're able to predict which words would come next in a sentence with high confidence, which allows them to generate language similar to how humans speak and write. These models are so advanced, in fact, that some have even questioned their ability to achieve sentience. But, while it's no secret that LLMs have become an important foundation for conversational AI systems, many people incorrectly assume that LLMs will eventually be the silver bullet that will solve all conversational AI problems -- and that's just not the case.

Microsoft plans to make OpenAI's generative AI technology such as ChatGPT available to billions of users by integrating it into all of its products, CEO Satya Nadella said this week. That means that ChatGPT's ability to generate text through short prompts is likely on its way to the Office 365 product suite, including Microsoft Word, PowerPoint and Outlook. Using OpenAI's models, Microsoft Word's autocomplete and autocorrect features could carry out more advanced tasks than style and grammar correction and generate longer chunks of text based on a few words. Though the company hasn't announced any specific features yet, users could potentially be able to input prompts and generate complete PowerPoint presentations and emails. These kinds of features could help Microsoft attract younger users. While Microsoft Office 365 has been a de facto standard for millions of enterprises, analysts say the tech giant has fallen behind in attracting those who gravitate toward collaborative-first products like Google Docs and Sheets.

Historically, security has been a binary rule-based system in which the state is either 0 (this file is benign) or 1 (congratulations, you have a virus). Complex systems define these classes based on a set of rules. But, how can you face this challenge at scale with more than 450,000 new malware types identified per day and more than 1.3 billion malware types already out there? How can Security Operations Center(SOC) teams handle the explosion in new types and the sheer scale of attacks? Furthermore, security experts need to understand why a file or an event is classified as malware or as an anomaly.

Artificial intelligence (AI) is helping prevent overfishing in a bid to protect the world's rapidly dwindling supply of edible marine species. A new project uses AI to improve the identification and measurement of fish species in Africa's Nile Basin. The software can help scientists understand fish population density more quickly than human observers. It's part of a larger effort to harness AI to improve sustainability across a wide range of industries. "The promising thing about AI is that it now allows us to do tasks that would be time-consuming or impossibly complex using traditional methods, with considerably more speed and efficiency," Andrew Dunckelman, head of impact and insights at Google.org, the search giant's charitable arm, told Lifewire in an email interview.

The "race" for dominance in AI is far from fair: Not only do a few developed economies possess certain material advantages right from the start, they also set the rules. They have an advantage in research and development, and possess a skilled workforce as well as wealth to invest in AI. We can also look at the state of inequity in AI in terms of governance: How "tech fluent" are policymakers in developing and underdeveloped countries? What barriers do they face in crafting regulations and industrial policy? Are they sufficiently represented and empowered at the international bodies that set rules and standards on AI?

Artificial intelligence (AI) is revolutionizing many processes across industries and applications--digital customer service assistants, autonomous vehicles, robots in retail warehouses. AI can even write a news article start to finish. There's certainly no lack of hype around the technology, and its application in business settings from a practical sense is nothing short of life-changing. However, AI comes with its drawbacks. It's not a silver bullet for every business qualm, despite what some hype men may promise.