Introducing GitHub Models: A new generation of AI engineers building on GitHub
[ad_1]
|
We believe that any developer with the right tools and training can be an AI engineer. From the playground to coding with the model in Codespaces to production deployment via Azure, GitHub Models shows you how easy it can be. Sign up for the limited public beta HERE. |
Since the dawn of the home computer, the dominant form of creation for developers has long been to build, customize, and deploy software with code. Today, in the age of AI, a secondary and equally important form of creation is rapidly emerging: the ability to use machine learning models. Developers are increasingly building generative AI applications, where the full stack includes backend and frontend code, as well as one or more models. However, a large portion of developers still lack easy access to open and closed models. This is changing today.
We’re launching GitHub Models, enabling our 100 million+ developers to become AI engineers and build with industry-leading AI models.
From Llama 3.1 to GPT-4o and GPT-4o mini to Phi 3 or Mistral Large 2, you can access each model through a built-in playground that lets you test different prompts and model parameters for free right in GitHub. And if you like what you see in the playground, we’ve created a glide path to bring the models into your developer environment in Codespaces and VS Code. And when you’re ready to go into production, Azure AI offers built-in responsible AI, enterprise-grade security and privacy, and global availability with provisioned throughput and availability in over 25 Azure regions for some models. Developing and running your AI application has never been easier.
The fun begins in the model playground on GitHub
For most of us, learning to become a developer was not a linear process in the classroom. It required practice, playing around, and learning through experimentation. The same is true today for AI models. The new interactive model playground allows students, hobbyists, startups, and more to test the most popular private And Open models from Meta, Mistral, Azure OpenAI Service, Microsoft, and others with just a few clicks and keystrokes. Experiment, compare, test, and deploy AI applications right where you manage your source code.
In keeping with GitHub and Microsoft’s ongoing commitment to privacy and security, no prompts or output in GitHub models are shared with model providers or used to train or improve the models.
|
Professor David J. Malan will test GitHub Models this fall at Harvard University’s CS50 to make it even easier for students to experiment with AI. |
Test and compare different models
Each software is unique. And likewise, each model is unique in its capabilities, performance and cost. Mistral offers low latency, while GPT-4o is great for building multimodal applications that require real-time audio, image and text. Some advanced scenarios may require the integration of different modes, such as an embedding model for Advanced Retrieval Generation (RAG).
With the model suite, developers have all the options they need to stay in the flow, experiment more, and learn faster than ever before. And this is just the first wave. In the coming months, as we move closer to general availability of GitHub Models, we will continue to add more languages, visions, and other models to our platform.
Bring your ideas to life with Codespaces
The fun isn’t just limited to the playground. Using the power of Codespaces, we’ve created a frictionless way for you to experiment with the model inference code before putting it into your own project. With ready-to-use example code for a variety of languages and frameworks of all kinds, you can try out different scenarios without ever running into “works on my computer” issues.
Then, when you’re ready, getting things up and running in your project is a breeze. Use the knowledge you’ve gained from the Playground and Codespaces to set up a prototype or proof-of-concept in your own applications. Run prompt evaluations in GitHub Actions with a series of JSON files that you can simply plug into the GitHub Models command within the GitHub Command Line InterfaceOr you can use GitHub Models to create a GitHub Copilot extensionextending GitHub’s platform ecosystem for every stage of software development. Finally, you can go into production with Azure AI by replacing your personal GitHub access token with an Azure subscription and credentials.
As an AI startup founder and open source maintainer, GitHub Models allows my team to access and experiment with different LLMs in one place, streamlining our development and lowering the barrier to entry for building AI apps.
The creator network for the age of AI
From developing artificial intelligence through open source collaboration, to building software with the power of artificial intelligence, to enabling the rise of the AI engineer with GitHub Models, GitHub is the developer network for the age of artificial intelligence.
The path to artificial general intelligence (AGI) is not possible without the source code and collaboration of the connected community on GitHub. Last year alone, more than 100,000 generative AI projects were created on GitHub.
GitHub Copilot is fundamentally changing the speed of software production and already writes almost 50% of the code into files when enabled. With GitHub Copilot WorkspaceWe envision a world where millions of beginners, hobbyists, and professional developers alike can program in a purely human language.
And now with GitHub Models, more than 100 million developers can access and experiment with new AI models where they already manage their source code, issues, pull requests, workflows, and repositories – right on GitHub.
In the coming years, we will further democratize access to AI technologies to reach one billion developers, empowering 10% of the world’s population to develop and drive breakthroughs that accelerate human progress for all.
Today we launch the limited public beta for GitHub Models. Register now. We can’t wait to see what you will build here.
Written by
[ad_2]
Source link