Code Llama

Code Llama is Meta's family of open-source large language models specialized for code generation and understanding, built on top of the Llama 2 architecture. Available in 7B, 13B, 34B, and 70B parameter sizes, with specialized variants for Python and instruction-following, Code Llama supports infilling, code completion, and natural language to code translation. The model handles up to 100K tokens of context and is released under a permissive license for both research and commercial applications.

Threads