Meta releases new programming aid model: CodeLlama70B

CodeLlama-70B-Instruct scored 67.8 on HumanEval, making it one of the highest performing open models available today.

Code Llama 70B is available in three versions:

  • CodeLlama – 70B, basic code model;
  • CodeLlama – 70B – Python, specifically for Python;
  • And Code Llama – 70B – Instruct 70B, fine-tuned for understanding natural language instructions.

Code Llama features include:

  • Exceeding the state-of-the-art LLM currently publicly available in coding tasks.
  • Free for research and commercial use.
  • Supports several popular programming languages, including Python, C++, Java, PHP, TypeScript (JavaScript), C #, and Bash.
  • We provide four models of different sizes (7B, 13B, 34B, 70B), each trained on a large amount of code and code-related data.
  • The model is able to provide stable generation, handling contexts up to 100,000 tokens.
  • Supports tasks such as code completion that are inserted into existing code.

Details: https://ai.meta.com/blog/code-llama-large-language-model-coding/
The content in this video has been automatically translated by safari

Video:

Scroll to Top