The Silicon Valley giant has released the first two models of the Llama 3 family – one with 8B parameters and one with 70B. According to Meta, these models are significantly better than the Llama 2 models, offering a much lower false rejection rate, improved alignment, and greater diversity in model responses. The specific capabilities of the models such as, code generation and instruction following have also been greatly improved.

Llama 3 was pre-trained on more than 15T tokens from publicly available sources, therefore the training dataset of Llama 3 is seven times larger than the training dataset of Llama 2 and contains four times more code.

According to Meta, in developing Llama 3, a new human evaluation set for benchmarking was also developed that contains 1,800 prompts across 12 use cases. These include asking for advice, classification, answering a closed question, writing code, creative writing, extraction, answering an open question, reasoning, rewriting, summarizing, and more.

The 70B parameter model beat Claude Sonnet, Mistral Medium, GPT 3.5, and Llama 2 using this new assessment set.

Llama 3 is available on AWS, Databricks, Google Cloud, Hugging Face, Kaggle, IBM WatsonX, Microsoft Azure, NVIDIA NIM and Snowflake. In addition, some hardware vendors will also offer support for it, including AMD, AWS, Dell, Intel, NVIDIA and Qualcomm.

Over the next few months, the company plans to update Llama 3 with new features, longer context windows, and larger model sizes.

Read more:
1. GitLab Releases GitLab Duo Chat with 40+ New Features
2. From Competitions in Informatics to MIT – Rumen Hristov’s Formula for Success
3. OpenSSF, CISA and DHS Join Forces in New Open Source Project

“The state-of-the-art open-access model improves on prior generative AI performance to increase developer productivity and provides developers equal access to the benefits of code generation AI, which in turn enables organizations of any size to more easily meet their full business potential”, said Harm de Vries, lead of ServiceNow’s StarCoder2 development team, and co-lead of BigCode.

StarCoder2 is trained on over 600 programming languages from The Stack v2 and some natural language code such as issues from Wikipedia, Arxiv and GitHub. The models use grouped query attention, a context window of 16,384 lexemes, with sliding window attention of 4,096 lexemes. Models 3B and 7B are trained on over 3 trillion lexemes, and model 15B on over 4 trillion lexemes.

According to the creators, the goal of StarCoder2 is to provide developers with features such as code generation, workflow generation, and text summarization, among others.

The project to develop StarCoder2 was led by ServiceNow and Hugging Face. StarCoder 2 is available in three distinct sizes: ServiceNow crafted a 3 billion-parameter model, Hugging Face developed a 7 billion-parameter model, and NVIDIA engineered a 15 billion-parameter model.

The smaller models are designed to offer powerful performance while using small amounts of compute power. According to the companies, the 3 billion-parameter model matches the performance of the 15 billion-parameter model of the original StarCoder release.

Read more:
1. Mark Zuckerberg Met with Japanese Prime Minister Fumio Kishida
2. Odysseus Mission Will Cease Operations
3. Meta Starts to Counter Disinformation

“The increasing complexity of today’s software ecosystems requires best-of-breed integrations between developer tools to help accelerate time to market without compromising security,” said Gal Marder, executive vice president of strategy at JFrog.

The latest JFrog Security feature integrated into Jira Cloud seamlessly incorporates JFrog security data into Jira, seamlessly weaving vulnerability management, application security, and compliance into developers’ daily workflows. This integration fosters improved collaboration and automation, guaranteeing reliable releases on a large scale. Currently, this feature is accessible in beta.

“To truly protect your software supply chain you need to consider code both in development and in production at the binary level. I look forward to further collaborating with our partners on solutions and go-to-market strategies that provide significant value to our customers wanting to migrate and innovate securely in the cloud”, Gal Marder added.

JFrog Workers, currently in open beta for JFrog SaaS customers, provides a serverless execution environment tailored for overseeing JFrog and third-party execution flows. This platform facilitates the development and execution of personalized scripts, adding an extra layer of security to automate and seamlessly interconnect developer workflows.

Additionally, it introduces new capabilities such as PagerDuty Security Incident Alerts through the integration of JFrog Xray with PagerDuty. Moreover, there are enhancements like Datadog Log Analytics and pre-configured log streaming for JFrog SaaS customers to Datadog and Splunk. These features are set to be available in open beta starting Q4 ’23.

Uno Platform serves as a foundation for constructing multi-platform .NET applications, enabling developers to streamline their workflow by utilizing a single codebase across diverse platforms.

Support for C#Markup has been added in the 5.0 release. According to the project’s maintainers, C# Markup has been desired by developers because it allows them to build a complete application using only this language.

“A massive bit of feedback from the community has been the need to learn multiple languages to build an Uno Platform application,” the Uno Platform Team explained in a blog post.

C# Markup demonstrates its versatility by offering support for a wide array of controls, including standard WinUI, Uno.Toolkit, and Uno.Extensions controls. Alternatively, developers can harness the C# Markup support generator to seamlessly integrate third-party controls. In addition, C# Markup is enriched with an array of powerful features such as data binding, styles, resources, templates, and visual states.

One of its standout advantages is the ability to create application user interfaces declaratively, establishing a clear demarcation between the UI and the underlying business logic.

Moreover, the Uno Platform Team has introduced addition: the Figma to C# Markup plugin, drawing inspiration from the Figma-to-XAML plugin. This innovative tool promises to further streamline the design and development process.

The team also created a course for learning about Hot Reload. The hands-on workshop guides developers through building a simple calculator that takes advantage of this feature. It also created another workshop that teaches how to build an app for streaming YouTube videos that can be deployed across multiple platforms, which also highlights the benefits of MVUX.

Other improvements in Uno Platform 5.0 include MP4 camera capture for iOS, improved composition support for Skia targets, improved DPI scaling and theming support for GTK, and more.

While the DevOps era has spawned tremendous innovation ranging from CI/CD, automated testing, and agile planning, the struggle to connect the innovations built through application development to the ways that customers experience those applications remains a challenge.

“The DevOps movement created years ago created an entirely new way to build software, yet there are still holes when it comes to the way customers interact with and experience applications today. With the Galaxy Product Release, LaunchDarkly hopes to provide engineering teams with a North Star that will solve for these gaps”, said Dan Rogers, CEO at LaunchDarkly.

The LaunchDarkly platform was built to guide engineers to the next frontier of DevOps by:

• Improving the velocity and stability of software releases, without the fear of end customer outages
• Delivering targeted experiences by easily personalizing features to customer cohorts
• Maximizing the business impact of every feature through the ability to experiment and optimize
• Coordinating the release and optimization of software to provide consistent experiences across mobile platforms and device types
• Improving the effectiveness and productivity of your engineering teams, by providing insights into engineering cadence and stability

“LaunchDarkly has helped JupiterOne’s software development teams unlock the power to release our software both faster and safer. We’re looking forward to diving into LaunchDarkly’s newly launched core capabilities into areas like product experimentation, release targeting, and measurement to see how they can continue to be such a successful component of our software delivery practices”, said DJ Spatoulas, Principal Architect at JupiterOne.

With this release, LaunchDarkly is continuing toward its vision of providing not just better developer experiences, but better customer experiences. The Galaxy Product Release includes:

• New capabilities to release stable features faster, including Release Assistant, to build pre-defined, repeatable paths for progressive rollouts, alongside Release Guardian, which allows teams to quickly identify and remediate operational regressions.
• Migration Assistant to reduce the risk of a migration such as an outage, data loss, or latency, while maintaining data consistency between data stores.
• Segment Builder to easily target customer groups with specific experiences through functionality that will sync and manage segments from existing data sources.
• Enhancements to LaunchDarkly’s Product Experimentation product, a new Funnel Experiments capability to measure and optimize customer behavior across full customer journeys. Mobile Release
• Optimization to liberate your mobile releases from app store processes and easily manage the full lifecycle of your apps.
• Engineering Insights Hub to help engineering leaders track velocity and quality metrics that show the measurable progress toward engineering transformation.

With this platform, businesses can tap into the capabilities of AI while adeptly managing the intricate challenges that organizations must conquer to deliver top-tier, secure software on a large scale.

“As companies embark on their AI adoption journey, we are seeing exponential improvements in application development. But with the vast adoption of AI code-assist tools, the question becomes, can DevSecOps processes, teams, and tools keep up with developer improvements? Businesses need to support an enhanced developer experience while overcoming roadblocks in their release pipelines, toolchains and security challenges. We have designed Denali to empower teams at every stage of the software development lifecycle (SDLC), helping to align developer outcomes with business strategy and accelerate innovation throughout the enterprise”, said Derek Holt, CEO of Digital.ai.

Denali guarantees seamless compatibility with the latest operating systems and development frameworks, elevating the performance of critical applications to new heights. Organizations can harness the potential of AI for enhanced automation of software delivery, effectively orchestrating and governing code originating from AI-assisted development. Moreover, it provides invaluable insights throughout each stage of the software delivery process, saving developers precious time and improving their access to knowledge.

“Our partnership with Digital.ai is focused on enabling secure digital transformation at leading
financial services companies. Ongoing R&D is crucial for us, and Digital.ai continually reinvests in their solution so that together, we can meet the fast-evolving needs of banking customers and help them deliver innovative applications that work for their customers. The new ARM Protection feature is an example of how Digital.ai makes application protection significantly easier while also eliminating extra steps”, said Jhonny Telles, Leadcomm’s Director of Digital Transformation.

Denali introduces a host of enhancements, including self-guided workflows, templates, and best
practices, streamlining processes and fostering alignment with quantifiable objectives. Furthermore, Denali extends its support for cloud-native application development and boasts enhanced integrations with prominent platforms like Terraform by HashiCorp, Azure Biceps, Azure Key Vault, and AWS Secret Manager.

In a post on the new social media app Threads, Meta’s owner said he had suggested the clash take place on August 26.

In a post on X (ex. Twitter), Elon says he’s been training “all day”, but Zuckerberg was quick to refute his claims, writing that Mr. Musk had not yet confirmed.

The two became direct competitors in July with the release of Threads.

Musk cast further doubt over the upcoming fight, saying the exact date was still in flux. He also said he is likely to need surgery on his neck and upper back.

The two social media moguls have not stopped challenging each other in recent months. On Sunday, Elon said their fight would be streamed live on his social media platform X.

He reaffirmed that all proceeds from the match will go to veterans charities.

The use of video calling apps has grown over the past few years, especially during the pandemic when telecommuting became a mainstream model and communication between people became online.

The company intends to extend the beta to Android developers later this year.

Interested developers, regardless of size, can sign up for the beta through T-Mobile’s DevEdge developer platform. Developers in the Seattle area will have the added benefit of joining T-Mobile engineers at the 5G Hub to test and validate this cutting-edge capability in their applications.

T-Mobile plans to expand network coverage to 300 million people by the end of the year.

Microsoft President Brad Smith is due to meet UK Chancellor Jeremy Hunt this week to discuss the CMA’s decision to block the purchase.

Microsoft President Smith is also expected to meet with Microsoft’s legal team to discuss any counter strategies. This meeting is also likely to discuss the ‘final’ option of withdrawing Activision Blizzard titles from the UK market.

“One of the extreme options could be to bypass the UK order and proceed with the deal or to withdraw Activision from the UK market.”

the report said.

Microsoft is expected to consider the EU’s findings in approving the purchase. In its ruling, the EU said the deal could “kick-start” competition in cloud gaming and streaming. However, the CMA said the deal would give Microsoft an unfair advantage over other companies such as Sony or Nintendo

Aside from Gears 5, there are plenty of other Xbox Game Studios and Bethesda titles coming to GeForce Now. Deathloop, Grounded, and Pentiment will be released on May 25 (Thursday). Nvidia says more Xbox games will be added regularly.

Although the first Xbox PC games will need to be purchased from Steam to be available on GeForce Now streaming, Microsoft and Nvidia are working on support for the Microsoft Store. It’s unclear if this will include PC Game Pass support. Nvidia also says that support for this store should “become available in the coming months”.

As you already know, earlier this year Microsoft struck a 10-year deal with Nvidia to license Xbox PC games for GeForce Now as part of a broader effort to appease European regulators over Activision Blizzard’s proposed deal. The EU approved the deal recently in the wake of similar 10-year deals offered to other cloud competitors. The deals also include access to Activision Blizzard titles if Microsoft’s proposed acquisition is approved by regulators.

If the giant deal with Microsoft falls through, Nvidia spokesperson Steffi Ngo claims to The Verge that the agreement with Microsoft and Xbox PC gaming will go ahead regardless of the outcome of the acquisition.