[Dev Catch Up #35] - Microsoft's Drasi and bitnet.cpp, next-safe-action library, Sana from Nvidia, MixEval-X framework, Integration test with Docker Compose model, and much more.

Bringing devs up to speed on the latest dev news from the trends including, a bunch of exciting developments and articles.

Welcome to the 35th edition of DevShorts, Dev Catch Up!

I write about developer stories and open source, partly from my work and experience interacting with people all over the globe.

Some recent issues from Dev Catch up:

• Ramp up with these Cursor resources, Reflection 70B

• Gemini Pro surpasses GPT-4o, Meta AI-Studio

• Meta unveils Llama 3.1, Open AI's GPT-4o mini

• LLM training at Meta, iOS 18 beta release, AI security with PCC

Join 1000+ developers to hear stories from Open source and technology

Must Read

• Microsoft unveiled a new open-source project, a data processing system named Drasi that simplifies detecting critical events within complex infrastructures and taking immediate actions tuned to business objectives. Learn more about Drasi from the official product blog.

• Microsoft made another banger of an announcement with the launch of bitnet.cpp. It is the official inference framework for 1-bit LLMs like BitNet b1.5B. Learn more about this open-source framework from its official GitHub repository.  

• Thumping their mark on the field of AI development, Meta announced the upcoming release of Catalina, the open architecture for AI infrastructure. This high-powered rack based on the Nvidia Blackwell platform is designed for handling AI workloads. Learn more about it from this article published by the Meta engineering team.

• Linters are static code analysis tools that can scan the codebase and flag potential problems and enforce coding standards. AST-Grep is one such tool that helps in doing that with the help of generative AI. Read this article to know more about it.

Now, we will head over to some of the news and articles that will be at a place of interest for developers and the tech community out there.

Good to know

• Meta Fundamental AI research team or FAIR has recently released a bunch of new researches, models, and datasets, which includes Segment Anything 2.1, Layer Skip, Spirit LM, etc. Learn more about these newly unveiled models, solutions, and datasets from this article from the FAIR team at Meta.

• The Testcontainers Docker Compose model is the solution for creating an integration test using multiple Testcontainers. Here is a detailed tutorial for the same.

• In React, the shadow DOM encapsulates the structure and styles of a web component which can limit external control. It is not very useful for system design and here is an article that highlights the potential future of web components without the need of it.

• Recently, Nvidia introduced a text-to-image framework that can synthesize and generate images efficiently with a 4096 x 4096 resolution at real high speed, deployable on laptop GPU. More information on it can be found here.

• An open-source library next-safe-action came into the limelight as it addressed the lack of essential features in powerful Server Actions in Next.js. Here is a complete guide to use it for Server Actions.

Lastly, we will take a look at some of the trending scoops that hold a special mention for the community. 

Notable FYIs

• Here is a pretty cool project named Manim that deserves recognition. It is an engine for precise programmatic animations, designed for creating explanatory math videos. Check it out from its official GitHub repository. Below is an image showing it in action.

  

• A new any-to-any, real-world benchmark named MixEval-X has been introduced which is designed to optimize and standardize evaluations across diverse input and output modalities. Learn more about this benchmark from here.

• Fork maintenance is a very common challenge that has been faced by every project maintainer. Here is a best practices guide that will help.

• Integrating LLMs in an application comes with a number of challenges and pitfalls. This advice and best practices guide will be a great help to tackle those challenges.

• Deepseek-ai launched their newly developed Janus-1.3B model which is constructed based on the DeepSeek-LLM-1.3b-base model and trained on approximately 500B text tokens. It can decouple visual encoding for multimodal understanding and generation. More information on the model is available on its huggingface page.

That’s it from us with this edition. We hope you are going away with a ton of new information. Lastly, share this newsletter with your colleagues and pals if you find it valuable and a subscription to the newsletter will be awesome if you are reading for the first time.