Search AI Products and News

Meta has introduced the all-new Meta LLM Compiler (Meta Large Language Model Compiler)! This compiler is built on the Meta Code Llama model series, available in both 7B and 13B versions. The Meta LLM features powerful code optimization and compiler capabilities; it can think like a compiler, predict the best paths for code optimization, and even disassemble code.

Meta has released Code Llama 70B, claiming it to be the best-performing open-source model. However, developers generally report that the high hardware costs are unaffordable. Code Llama performs exceptionally well in the HumanEval and MBPP tests, surpassing GPT-3.5 and Llama 2. Critics point out that using the 70B model may require expensive hardware, and some developers have indicated that it is unable to run on lower-spec machines. Meta states that the model supports multiple programming languages, including Python.

Meta has released the open-source LLM Code Llama, built on Llama 2, claiming superior performance on coding tasks compared to other LLMs. Code Llama comes in three variants: a base model, a model specifically for Python, and a model for natural language instructions. Code Llama provides stable generation and supports multiple programming languages, suitable for tasks such as code completion, debugging, and real-time code assistance.