Home » Meta releases powerful ‘Code Llama 70B’, an open-source coding tool to rival

Meta releases powerful ‘Code Llama 70B’, an open-source coding tool to rival

by OnverZe
1 comment

Meta has released Code Llama, an advanced AI model with 70 billion parameters. It’s available for free download for researchers and developers since its announcement in August 2023.

We are excited to announce that Code Llama, an AI-powered tool for writing and editing code, is now available as an open-source project. The latest version includes a larger 70B parameter model which will enable AI models to process information more logically and rigorously, making it one of the most important uses of AI models today. Meta CEO Mark Zuckerberg took to Facebook to express his pride in the progress made and anticipation for the upcoming release of Llama 3 and future models.

Meta claims that Code Llama is the most advanced and highest-performing model in the Llama family. It is available for free in three versions: CodeLlama – 70B, which is the foundational code model, CodeLlama – 70B – Python, which is 70B specialized for Python, and Code Llama – 70B – Instruct 70B, a version fine-tuned with human instruction and self-instruction code synthesis.

Code Llama is built on top of Llama and can generate code. According to the company, the model has achieved a score of 67.8 on HumanEval, which is a generative AI benchmark. The GPT-4 Turbo, which is a much larger model, has a score of 81.7 in the same benchmark. Meta also claims that Code Llama is specifically optimized for code generation and the best part is that it is an open-source model.

Meta has developed a new generative AI model called Code Llama that can write code in C++, Java, PHP, and Python. This model has been trained using a vast amount of data, comprising 500 billion tokens of code and related material. Meta has also expanded the context window, which can now accept inputs with up to 100,000 tokens or around 75,000 words.

 It is important to note that Code Llama is based on Llama, a general-purpose AI model with 175 billion parameters. Users can access Code Llama through platforms like Hugging Face or PyTorch, and the company has confirmed that some of these advancements will be included in the upcoming Llama 3 and future models by Meta.

You may also like

1 comment

Leave a Comment