Running your personal coding copilot locally

Why Spend $10 on GitHub Copilot When You Can Run Your Personal Copilot Locally?

In the realm of coding, efficiency, and support tools can make a world of difference. That's why I embarked on an exploratory journey to leverage open Large Language Models (LLMs) locally, bypassing the need for GitHub's Copilot subscription. The outcome? A suite of features that not only rivals but in some aspects, surpasses the convenience of Copilot - all without the need for internet access.

🚀 Features Unleashed Locally:

👉 Autocomplete for Python & Kubernetes YAMLs: Plus, support for other programming languages.
👉 Code Explanation for Java: Making sense of complex snippets has never been easier.
👉 Interactive Enhancements: Ask follow-up questions to refine and improve your code.
👉 Unit Test Generation: Automate the creation of essential tests.

🛠 Models at Play:
👉stable-code:3b-code-q4_0
👉deepseek-coder:1.3b-base-q4_0
👉codellama:7b-code-q4_0

🔌 Free IDE Extensions Showcased:

While the results are still catching up to GitHub Copilot's benchmark, the rapid improvement of open-source models is undeniable. What's more, these IDE plugins can be directed towards OpenAPI models using a personal API key for those requiring internet-based features.

The journey into leveraging local LLMs for coding has been a revelation, showcasing the potential for a more personalized and cost-effective coding assistant. The future of coding looks bright, with open-source innovations leading the charge.

Happy Coding!