Meta releases AI developer assistant ‘Code Llama’

Facebook parent Meta announced on Aug. 24 that it is releasing Code Llama, a large language model (LLM) for use by coders and developers.

The company said that Code Llama can generate and discuss code, improve workflow speeds, and reduce barriers to entry for new coders.

Meta added that the new product is a code-specialized version of Llama 2, its open-source large language model. The new product is available for research and commercial use under the same license as Llama 2, which was opened in July.

Meta claims that Code Llama can generate code as well as natural language about code, and that the tool can also accept both types of information as input. The company also said that, in addition to generating code and content, Code Llama can be used for debugging purposes. Meta said that it supports popular programming languages, including Python, C++, Java, PHP, Typescript (Javascript), C#, Bash, and others.

The company said that it is releasing three Code Llama versions, each trained on different parameters and requirements. The smallest 7B version can be served on a single GPU, according to Meta’s latest announcement. The largest 34B model, on the other hand, returns better results and offers better coding assistance.

The company also announced a Python-focused version of Code Llama, as well as an instruction-focused version of Code Llama that excels at understanding human prompts.

AI can be used by blockchain developers

Within the blockchain industry, some have suggested that AI can be used to help write smart contracts that underpin popular decentralized apps (dApps) and DeFi platforms — and to discover vulnerabilities in existing smart contracts.

Though Meta did not announce support for popular smart contract languages such as Solidity, other projects such as ChainGPT and Decrypted are working to provide similar features that allow developers to generate such code through an AI tool.

The post Meta releases AI developer assistant ‘Code Llama’ appeared first on CryptoSlate.

Credit: Source link