IBM Granite for Code models are available on Hugging Face and ready to be used locally with “watsonx Code Assistant”

If you’re a developer or involved in coding, the IBM Granite for Code models on Hugging Face might be useful for you. To save you time searching for resources, I’ve compiled a brief post with some of the available resources.

IBM Granite for Code models, conveniently accessible on Hugging Face, can be seamlessly integrated into VS Code. This ease of integration empowers developers, making the models easy to consume and enhancing the coding experience.

When you use AI assistants in your coding, it’s important to reference them in your code documentation. This responsible practice can help you in the future, ensuring that your code is not mistaken for plagiarism. For more on this, check out the IBM article Impact on education: plagiarism risk for AI | IBM watsonx.

The GIF below shows a short preview of the integration in VS Code with watsonx Code Assistant.

Some highlights

• Trained for 116 programming languages
• Available under Apache 2.0 license for research and commercial use
• Available on Hugging Face for usage
• You can directly integrate them into VS Code
• Accessible playground for an easy and fast first interaction
• Usage for internal chat and code generation

Helpful information resources

• GitHub Granite Code Models
• Granite Play Ground on IBM.com
• Introduction articles from IBM Developer:
  • Introducing IBM Granite.Code – Unlocking the Power of the Granite large language model
  • Build a local AI co-pilot using IBM Granite Code, Ollama, and Continue
    
• VS Code Marketplace “IBM Granite.Code”
• Hugging Face Granite Code Models

I hope this was useful to you and let’s see what’s next?

Greetings,

Thomas

#vscode, #codemodel, #ai, #developer, #ibm, #ibmdeveloper, #granite, #granitecode, #ai, #watsonxcodeassistant, #codeassistant

PS: I am using the command line to start Ollama (the inference to run the models locally), and I have now installed “granite-code:20b 11GB” and “granite-code:8b 4.6 GB”.

/opt/homebrew/opt/ollama/bin/ollama serve