Code Llama, a new AI-powered tool that Meta has released amid the AI competition, will assist programmers and IT professionals in creating code and debugging human-written work. The new tool from Meta directly competes with ChatGPT, the most active AI model in OpenAI, which is presently assisting users with projects and programmes.
The free availability of Code Llama, an AI coding tool from Meta, was revealed on Thursday. can finish and debug code when directed to a specific code string, as well as construct strings of code from prompts.
The business underlines that customers will be able to utilise the Code Llama to make their workflows quicker and more effective as well as to decrease the entrance barrier for those who are learning to code. The potential for using Code Llama as a productivity and training tool to assist programmers in creating more robust, well-documented software exists.
with a blog post, Meta stated that “programmers are already using LLMs to help in a variety of tasks, ranging from writing new software to debugging existing code.” According to Meta’s official blog article, the objective is to increase developer workflow efficiency so they can concentrate on the parts of their professions that are most focused on people.
Meta asserts that Code Llama outperformed other publicly accessible language models in benchmark testing, although it doesn’t say which ones. The business added that Code Llama can accurately produce code based on a text description and achieved a score of 53.7% on the code benchmark HumanEval.
The smallest form of Code Llama, which Meta will distribute in three sizes, can run on a single GPU for low-latency tasks.
The AI coding tool, Code Llama, by Meta has been released in three different iterations. “We are making available three sizes of Code Llama, each containing 7B, 13B, and 34B characteristics. 500 billion tokens of code and data linked to code are used to train each of these models.
The Python-specific version is made for Python programming, but the basic version may be used for generic coding jobs. The natural language version can comprehend instructions given in plain English. For natural language instructions, Meta does not advise utilising either the basic version or the Python-specific version.
The open-source language model Llama is one of many AI models that Meta, led by Mark Zuckerberg, has released for free. The OpenAI, Microsoft, and Google proprietary models are seriously threatened by this development.
The new AI model, which will be offered in several variants, is based on Meta’s most current language model, Llama 2. This approach is being made available by Meta in order to compete with Microsoft’s code-generation tool GitHub Copilot.
“Code Llama is made to help software developers in any field, including open source initiatives, NGOs, corporations, and research. However, there are still a tonne of more use cases to support,” Meta stated.
1 comment
[…] Technology […]