Code Llama: Open Foundation Models for Code

Code Llama: New open models that write and finish code

Code Llama is a family of code-writing models made for people who build software and learn coding.
It comes in several sizes, and each are aimed at different users, some small and quick, some large and powerful.
There is a version tuned for Python, one that follows simple instructions, and models that can read very long files.
You can ask it to finish a function, or to fill a missing part in the middle of a file — that filling trick is called infilling and it works well even across long code files.

The makers say these models do better than other publicly available ones on many tasks, and they released them under an easy, permissive license so people and companies can use them.
Try it when you need faster drafts, clearer examples, or a quick hand with repetitive code.
It feels helpful, and it learns the style of your project, so over time it writes in tone that matches yours.
The models are open for use, play and building new tools.

Read article comprehensive review in Paperium.net:
Code Llama: Open Foundation Models for Code

🤖 This analysis and review was primarily generated and structured by an AI . The content is provided for informational and quick-review purposes.