Meta is back with yet another AI model to help developers. Code Llama is an AI model that is built on top of Meta’s Llama 2. The latest tool is meant to generate and discuss code and is free for research and commercial use. The state-of-the-art language model can generate codes based on text prompts. It has the potential to enhance workflows for developers and even make coding accessible for beginners.
“We believe an open approach to AI is best for developing new AI tools that are innovative, safe and responsible, so we’re releasing Code Llama for both research and commercial use under the same community license as Llama 2,” Meta said in a post on its website.
Most Read
Gadar 2 box office collection day 16: Sunny Deol film beats KGF 2, earns Rs 439.95 crore
World Athletics Championships 2023 Highlights: Neeraj Chopra wins gold; Relay runners finish 5th; Parul Chaudhary scripts national record
According to Meta, Code Llama is a code-specialised version of Llama 2 which was trained on code-specific data sets. The latest model comes with enhanced coding capabilities and can generate code and natural language about code from both code and prompts in natural language. Code Llama can also be used for code completion and debugging. It also supports programming languages such as Python, C++, Java, PHP, Typescript (Javascript), C#, Bash and more.
Meta has released Code Llama in three sizes with 7B, 13B, 34B parameters and each of these models has been trained on 500B tokens of code and code-related data. The 7B and 13B have been trained with fill-in-the-middle (FIM) capability which allows them to add code into existing code.
Code Llama is also equipped with specialised variations and instruction fine-tuning. While Code Llama – Python is fine-tuned on Python code offering additional utility for developers, the Code Llama – Instruct is fine-tuned on instruction input to generate helpful and safe answers in natural language.
/cloudfront-us-east-2.images.arcpublishing.com/reuters/3FYYYZWGPJLNDMW3GC3P3MRFOQ.jpg)
