Open Source News & Articles on Techminis

A naukri.com initiative

New

Home

Open Source News

Sdtimes

387

Image Credit: Sdtimes

ActiveState relaunching its platform for open source management

ActiveState, which previously managed open source dynamic programming languages, has rebranded and released an open source management platform to help users streamline DevSecOps, manage open source risks and ensure supply chain security with tools for discovery, analysis, remediation and governance.
It proactively manages open-source risks, and ensures reproducible builds and streamlines upgrades, reducing the burden on developers.
Most software applications today are less secure than they have ever been, Baker said.
Software supply chain attacks are expected to double by 2030, to about $140 billion. Organizations need to become much more proactive in how they manage open source, using tools to enforce policies that cause the least amount of disruption to the development process and foster greater collaboration, said ActiveState CEO, Stephen Baker.
Among the capabilities of new ActiveState Open Source Management Platform are the ability to discover open source, monitor it through a single dashboard, and facilitate analysis, prioritisation, policy and governance.
ActiveState integrates with existing tools to keep the open source current and more secure. It eliminates 90% of the undifferentiated heavy lifting that every developer needs to do to research dependencies, upgrades, and risk before integrating new open-source into their organization.
ActiveState offers an open source supply chain security platform that is built on automation to provide timely insights into how vulnerable open source is and what developers should do to make it less vulnerable.
The platform is for customers to manage their own open-source that they are consuming, a much more secure format that will improve the application security posture and, in return, not destroy developer productivity.
With ActiveState, there is no reactive model. Instead, it applies all the rules and policies even before open-source gets into user's organizations, so that clean and compliant open source is incorporated when the application is being built.
Every organization on the planet is now dependent on open source, and threat actors and cyber attackers are now depending on the lack of organizational controls on open source to plan their next attack.

Read Full Article

23 Likes

Siliconangle

Image Credit: Siliconangle

Red Hat to acquire AI optimization startup Neural Magic

Red Hat Inc. plans to acquire AI optimization startup Neural Magic.
Neural Magic specializes in optimizing AI models to run on different hardware, including CPUs and GPUs.
The acquisition will enhance Red Hat's ability to support AI deployments across hybrid cloud environments.
Neural Magic's expertise in model deployment and the open-source project vLLM played a key role in the acquisition.

Read Full Article

26 Likes

Siliconangle

797

Image Credit: Siliconangle

LocalStack nabs $25M investment for its public cloud emulator

Swiss developer tooling provider LocalStack GmbH has raised $25 million in an early-stage funding round.
The funding round was led by Notable Capital, with participation from CRV and Heavybit.
LocalStack offers an open-source tool that allows developers to emulate an Amazon Web Services environment on their local machines.
The platform helps developers test cloud applications, infrastructure configurations, and enables chaos engineering and database migration testing.

Read Full Article

8 Likes

Medium

Image Credit: Medium

Introducing MonitorFlow: An Open-Source SaaS for Real-Time Monitoring and Analytics

MonitorFlow is an open-source SaaS for real-time monitoring and analytics.
It offers a free, flexible, and extensible solution for monitoring applications and infrastructure.
MonitorFlow brings exciting features and invites contributions from the community.
Contributors can gain hands-on open source experience and make a real impact on shaping MonitorFlow.

Read Full Article

4 Likes

VentureBeat

Image Credit: VentureBeat

Qwen2.5-Coder just changed the game for AI programming—and it’s free

Alibaba Cloud has released Qwen2.5-Coder, a new AI coding assistant.
The release includes six model variants, making advanced AI coding accessible to developers with different computing resources.
Qwen2.5-Coder's success stems from refined data processing, synthetic data generation, and balanced training datasets.
Unlike its closed-source competitors, most Qwen2.5-Coder models carry the permissive Apache 2.0 license, allowing free integration into products.

Read Full Article

3 Likes

TechCrunch

194

Image Credit: TechCrunch

Dapr graduates to become a CNCF top-level project

Dapr, the open-source runtime for building distributed applications, has graduated from the CNCF's incubating projects to become a top-level project.
To reach this level, a project must be stable, well-documented, have active maintainers, and gain market traction.
Dapr aims to simplify distributed application development and has been supported by over 3,700 contributors from 400 organizations, with tens of thousands of users.
Future plans for Dapr include adding a conversational AI API in the next release to enable developers to work with various language models.

Read Full Article

11 Likes

Marktechpost

221

Qwen Open Sources the Powerful, Diverse, and Practical Qwen2.5-Coder Series (0.5B/1.5B/3B/7B/14B/32B)

Qwen has open-sourced the Qwen2.5-Coder series, dedicated to promoting open CodeLLMs.
The Qwen2.5-Coder series offers models ranging from 0.5B to 32B parameters, providing flexibility for coding needs.
Qwen2.5-Coder has achieved state-of-the-art performance in multiple evaluation benchmarks and excels in multi-programming language capabilities.
The availability of models in various parameter sizes ensures scalability and accessibility for developers.

Read Full Article

13 Likes

VentureBeat

Image Credit: VentureBeat

Google DeepMind open-sources AlphaFold 3, ushering in a new era for drug discovery and molecular biology

Google DeepMind has released the source code and model weights of AlphaFold 3 for academic use, advancing scientific discovery and drug development.
AlphaFold 3 can model the interactions between proteins, DNA, RNA, and small molecules, transforming it into a comprehensive solution for studying molecular biology.
The release balances open science and commercial interests, with freely available code and controlled access to model weights.
AlphaFold 3's technical advances and improved accuracy in predicting molecular interactions have implications for drug discovery and other fields.

Read Full Article

4 Likes

Medium

447

AIM Weekly #163 for 11 Nov 2024

Synthetic Data Generator — local on Ollama!
Convert an entire HTML Website to Markdown with Great GO CLI Tool
Creating Advanced AI Agents with Ollama and Langchain
Tencent’s Supercharged MoE Model

Read Full Article

26 Likes

Marktechpost

167

Hugging Face Releases Sentence Transformers v3.3.0: A Major Leap for NLP Efficiency

Hugging Face has released Sentence Transformers v3.3.0, a major update with significant advancements especially for efficiency and usability for a broader audience.
The latest version is packed with features that address performance bottlenecks, enhance usability, and offer new training paradigms.
This latest version has a 4.5x speedup for CPU inference by integrating OpenVINO’s int8 static quantization.
The integration of OpenVINO Post-Training Static Quantization allows models to run 4.78 times faster on CPUs with no performance drop.
The introduction of training with prompts improves the performance in retrieval tasks by 0.66% to 0.90% without any additional computational overhead.
PEFT integration allows for more scalability in training and deploying models reducing memory requirements.
The ability to evaluate on NanoBEIR adds an extra layer of assurance that the models trained using v3.3.0 can generalize well across diverse tasks.
The release shows Hugging Face’s commitment to enhancing computational efficiency, making these models more accessible across a wide range of use cases.
This update ticks all the right boxes for developers, ensuring models are not just powerful but also efficient, versatile, and easier to integrate into various deployment scenarios.

Read Full Article

10 Likes

TechCrunch

285

Image Credit: TechCrunch

Open source projects draw equity-free funding from corporates, startups, and even VCs

Various sector players are stepping up with funding initiatives to support open-source tech including startups, corporations, and venture capital firms.
While some open source software has the potential to be commercialised, many projects are difficult to monetise.
Funding programmes include reactive initiatives such as the $30m pledge launched in response to the Log4Shell security flaw, and proactive offerings driven by various industry giants.
Sequoia Capital has launched an open source fellowship to support project maintainers with equity-free capital of up to $100,000 for up to 12 months, while Andreessen Horowitz launched an open source AI grant program in 2020.
Chatbot Arena and vLLM are two projects selected for Sequoia’s fellowship. Both were birthed out of UC Berkeley’s Sky Computing Lab and have PhD researchers working on them. These were funded by Sequoia at $100K each for this year.
Founder of AI Factory, Eno Reyes, has confirmed that his company uses Chatbot Arena to keep track of the top LLM options.
Sentry, a developer tooling unicorn, has launched the Open Source Pledge to encourage companies to get involved in donating to open source projects.
The Open Source Pledge calls for contributions of at least $2,000 per year for each developer on staff.
Sentry has committed $750,000 this year and has over two dozen members signed up to the Pledge at the time of writing.
Accel partner Dan Levine said, The open source community, particularly on the demand side, needs to reassess its strategies and make more informed decisions to support critical projects".

Read Full Article

17 Likes

Medium

240

Image Credit: Medium

Bridging the Gap: Hugging Face Transformers and GitHub for Cross-Framework NLP

The Hugging Face Transformers library and GitHub have revolutionized NLP by democratizing access to transformer models.
The Hugging Face Model Hub on GitHub serves as a central repository for sharing and collaborating on these models.
Two case studies showcase the flexibility and ease of using Hugging Face Transformers in PyTorch and TensorFlow for text generation and summarization.
The framework-agnostic nature of the library and the collaborative features of GitHub promote accessible and community-driven NLP development.

Read Full Article

14 Likes

Medium

335

Image Credit: Medium

Mastering Git (Part 2): Going Beyond the Basics

In this article, we’ll take you beyond the basics and dive into advanced Git.
Git’s branching mechanism allows you to work on different features or fixes in isolation, without disrupting the main codebase.
There are two main ways to create and switch branches in Git.
Once you’ve made your changes on a branch, it’s time to merge them back into the main branch.
Merge conflicts are like the “awkward dinner conversations” of Git.
When you’re working solo or collaborating, remote repositories are your best friend.
Rebasing is a powerful tool, but it’s like the “nuclear option” for commit history — use with caution!
Ever realized you made a small mistake right after committing? Don’t worry — you don’t have to create a whole new commit to fix it. Git lets you amend the last commit.
Git is powerful, but some commands can be long and cumbersome. Instead of typing out the full command every time, you can alias them to something shorter and more memorable.
Working with Git in a team environment requires some extra care to keep everything organized.

Read Full Article

20 Likes

Schneier

Image Credit: Schneier

AI Industry is Trying to Subvert the Definition of “Open Source AI”

The Open Source Initiative has published its definition of “open source AI,” which allows for secret training data and mechanisms.
Many “open source” AI models are open source in name only, and industry players want both corporate secrecy and the “open source” label.
There is a need for a public AI option, and real open source is necessary for that.
The exclusion of some training data is allowed in fields where data cannot be legally shared, such as medical AI.

Read Full Article

2 Likes

Analyticsindiamag

117

Image Credit: Analyticsindiamag

IBM Commits to Transparency with Granite 3.0 LLM

IBM has released new open-source model, Granite 3.0. The datasets that are used for training are clearly disclosed by IBM. The language models are trained on Blue Vela, powered entirely by renewable energy.
Recently, Stanford’s Foundation Model Transparency Index report stated that IBM’s Granite models achieved a 100% score in transparency and openness of a model test.
Big tech companies including Google, Apple, and OpenAI remain tight-lipped about the data used to train their LLMs.
IBM’s approach is to increase transparency by default. It can help businesses overcome the scepticism and negative ‘black box’ connotation attached to LLMs.
By providing transparent models, IBM is likely to receive more recognition for playing the good guy in the world of AI.
Deep integration of AI into technology is something that people will choose on how to use or if they want to use it.
Granite models aim to make open source the winner in the world of the internet. People encountering good samaritans may get motivation to contribute to open source datasets.
IBM’s generative AI book of business is up by more than $1 billion quarter to quarter.
IBM's transparency can help businesses spend more time looking for solutions to their problems rather worry about the reliability of models they’re using.
IBM’s training datasets are mainly open-source, which would help in research, development and enhance these datasets.

Read Full Article

7 Likes

For uninterrupted reading, download the app

Open Source News

Discover more