A critical part of creating code suggestion systems is the pre-training of Large Language Models on vast amounts of source code and natural language text, often of questionable origin or quality. This may contribute to the presence of bugs and vulnerabilities in code generated by LLMs. While efforts to identify bugs at or after code generation exist, it is preferable to pre-train or fine-tune LLMs on curated, high-quality, and compliant datasets. The need for vast amounts of training data necessitates that such curation be automated, minimizing human intervention. We propose an automated source code autocuration technique that leverages the complete version history of open-source software projects to improve the quality of training data. This approach leverages the version history of all OSS projects to identify training data samples that have been modified or have undergone changes in at least one OSS project, and pinpoint a subset of samples that include fixes for bugs or vulnerabilities. We evaluate this method using The Stack v2 dataset, and find that 17% of the code versions in the dataset have newer versions, with 17% of those representing bug fixes, including 2.36% addressing known CVEs. The deduplicated version of Stack v2 still includes blobs vulnerable to 6,947 known CVEs. Furthermore, 58% of the blobs in the dataset were never modified after creation, suggesting they likely represent software with minimal or no use. Misidentified blob origins present an additional challenge, as they lead to the inclusion of non-permissively licensed code, raising serious compliance concerns. By addressing these issues, the training of new models can avoid perpetuating buggy code patterns or license violations. We expect our results to inspire process improvements for automated data curation, with the potential to enhance the reliability of outputs generated by AI tools.

Robots, and in particular industrial robots, are programmed to perform certain functions. The Robot Operating System (ROS) is a very popular framework that facilitates the asynchronous coordination between a robot and other drives and/or devices. ROS has been a go-to means to enable the development of advanced capability across the robotics sector. Southwest Research Institute (SwRI) and the ROS-I community often develop applications in ROS 2, the successor to ROS 1. In many cases, particularly where legacy application code is utilized bridging back to ROS 1 is still very common, and one of the challenges in supporting the adoption of ROS for industry.

The Beijing Academy of Artificial Intelligence (BAAI) unveiled a newer version of its hyper-scale pre-trained deep learning model, the country's first and the world's largest, at an ongoing AI-themed forum in Beijing, in the latest signal of China's ambition to become a global leader in AI. The latest version of the model, known as Wudao, literally meaning an understanding of natural laws, sports 1.75 trillion parameters, breaking the record of 1.6 trillion previously set by Google's Switch Transformer AI language model, the academy announced Tuesday at the three-day forum that runs through Thursday. Wudao was only initially released in March. Wudao is intended to create cognitive intelligence dually driven by data and knowledge, making machines think like humans and enabling machine cognitive abilities to pass the Turing test, Tang Jie, BAAI's vice director of academics, said during the forum. The newer version of Wudao is both gigantic and smart, featuring its hyper scale, high precision and efficiency.

Neuralink, Elon Musk's neurotechnology startup that is developing interfaces to connect human minds to computers, held a big event yesterday to announce what progress they had made over the last one year. For the uninitiated, Neuralink aims to help people with severe neurological disabilities via devices that can be implanted in the brain. Neurons, which are the basic working units of the nervous system, transmit signals to other parts of the body. All of that information exchange happens over minuscule electrical signals. With the Neuralink device in place, patients should be able to regain or improve upon the effects of neurological injuries.

If you are a long time Oracle customer, you may recall the days in 2009 when Oracle 11g along with Oracle Enterprise Manager 11g was released. Oracle Enterprise Manager Patching and Provisioning Pack 11g included capabilities for automated database provisioning, patching and upgrade. Compliance for Oracle recommended Critical Patch Updates (security patches) was achieved when Patch Advisories in Enterprise Manager first discovered patch drift and then provided automated mechanisms (deployment procedures) to correct the drift. A decade later, Oracle has transformed. Oracle's Cloud strategy and unmatched breadth of business applications provides a unique opportunity for utilizing data to make AI and Machine Learning fulfill its promise.

This is a code repository for the upcoming book "Java Deep Learning cookbook" sponsored by Packt Publishing. We use and promote deeplearning4j library for all use-cases in this book. Official deeplearning4j version targeted in this cookbook is 1.0.0-beta3. For the same reason, some of the methods or approaches discussed in this cookbook may get deprecated in their newer versions. So, be sure to refer their latest API documentation.

In 2001, the video-game studio Jagex launched Runescape, an online role-playing game. This week, the developers announced that the original version of the game, one that consumed vast quantities of my time as a middle schooler and remained online until now, is shutting down. In 8th grade I desperately wanted to play Everquest, the dominant online role playing game at the time, but I didn't have a credit card, and my parents wouldn't let me use theirs. So I typed in "free online RPG" into a search bar and my brief, fiery affair with Runescape began. I played it in a browser, and it was archaic looking even then compared to Everquest's graphics.

In its day two keynote for Build 2018, Microsoft showed off a brief demonstration of how it is planning to bring the power of machine learning to Word. In a future version of the office app, Microsoft will tap machine learning to help writers improve their grammar. Microsoft was pretty light on details of what to expect, but what it showed off looked promising. On stage, Kevin Gallo, head of the Windows developer platform, displayed an older version of Word compared to the newer version with a machine learning-powered grammar checker built in. While the older version missed several grammar issues, the newer version found them all.

Mojang, the developer behind popular video game Minecraft, has announced that after the upcoming'Update Aquatic' release, the game will not update on older consoles. Players on PlayStation 3, Xbox 360, PS Vita and Wii U will all receive the latest update shortly after the initial launch hits newer systems. But this will be the last new update that is compatible with these platforms. 'This will mark the last big effort to keep those editions up to date with our newer versions of the game,' Mojang said in a post on their website. 'Although we'd love to keep bringing new content to all our players forever, the older generation of consoles now make up less than 5 percent of our active players' 'So we've made the difficult decision to focus our efforts to support players where they play Minecraft the most and where we plan to add new features,' they added.

If you love playing Minecraft on the PS3, Xbox 360, Wii U or PS Vita, the game's developers have some bad news for you. The next update, called "Update Aquatic," will be the last that is compatible with these platforms. Only about 5 percent of active players use Minecraft on these older consoles, and the team has chosen to put its resources where the majority of people who play the game are. The PS4 will continue to receive updates, as will mobile and Windows 10 versions. The Xbox One and Switch versions of the older game will not receive the Update Aquatic because users can upgrade to the newer version.