Meta, creators of the leading social networks of the moment, unveiled the creation of a scripting tool for programming called Code Llama. Its function is to use text to generate code thanks to artificial intelligence (AI).

Code Llama can be used as an educational tool to help programmers write more robust and well-documented software. You will also be able to generate code and natural language over the code.

Code, META’s new tool, helps businesses and entrepreneurs innovate, benefiting society as a whole.

It supports the following programming languages: Python, C++, Java, PHP, Typescript (Javascript), C#, Bash and more. There will be three sizes of Code Llama with parameters 7B, 13B and 34B respectively.

Two variations of Code Llama were created: Code Llama – Python and Code Llama – Instruct.

Code Llama – Python, is an optimized variation on over 100 billion tokens of Python code. This variation was created because Python has a prominent role in the AI community and can give extra utility to developers.

Code Llama -Instruct, is a tuned and aligned instruction variation of Code Llama.

In this model, an instruction input is received in natural language and gives the expected result. Helping to give useful answers in natural language which improves the outcome the person expects.

In the press release Meta highlights that this tool is free for commercial and research use. He indicates that they have a wide range of companies from different sectors who are supporting them and who, like them, believe in an open approach to new AI technologies.

For this reason, it urges responsible use of its tool and will provide resources for those who use the tool to do so as well.

To achieve this they are working on:

Red-Teaming exercises: whose models have undergone extensive testing to facilitate model tuning, both internally and externally, and have also included third parties for performance testing. The idea is to publish more polished models with greater security enhancements.

Responsibility guide: they have created a guide to help developers keep development best practices in mind.

Acceptable use policy: this policy prohibits and establishes cases where its use will not be possible and ensures that these models are used in a responsible and fair manner.

Transparency scheme: they seek to explain the shortcomings of the system and the challenges and drawbacks they have experienced, providing information on the measures they have taken to improve it.

Meta highlights that “generative AI has the potential to help people do amazing things, create a new era of economic and social opportunity, and give individuals, creators and businesses new ways to express themselves and connect with people.”

Code llama as META’s new programming tool, seeks to benefit all businesses, companies and entrepreneurs, with the support of this powerful tool, which is accessible to all and will allow them to create and innovate in their sectors, which will ultimately benefit the whole society.

Microsoft continues its collaboration agreement with Meta for the Llama 2 repurposing and will put it in its catalog of Azure AI models.

Such an alliance will enable developers using Microsoft Azure to build and leverage these cloud-native tools for content filtering and security features.

But this partnership has been extended to the long-awaited metaverse where it will offer immersive experiences for the future of work and play.

Llama 2 has the advantage that it can run locally on Windows, so that developers can work seamlessly across platforms.

Amazon Web Services (AWS), Hugging Face and other providers will have Llama 2 available.

Source: Meta news

If you are looking for a company to help you get into the world of artificial intelligence, let’s talk, we are here to help you.