If you’re interested in using generative A.I. tools to write and debug code, you have another option: Meta’s just-released Code Llama.
“Code Llama is a code-specialized version of Llama 2 that was created by further training Llama 2 on its code-specific datasets, sampling more data from that same dataset for longer,” is how Meta described the tool in a new corporate blog posting. “Essentially, Code Llama features enhanced coding capabilities. It can generate code and natural language about code, from both code and natural language prompts (e.g., ‘Write me a function that outputs the fibonacci sequence’).”
Meta is releasing Code Llama in 7B, 13B, and 34B parameters. “Each of these models is trained with 500B tokens of code and code-related data,” the posting added. “The 7B and 13B base and instruct models have also been trained with fill-in-the-middle (FIM) capability, allowing them to insert code into existing code, meaning they can support tasks like code completion right out of the box.”
Those developers who want maximum A.I. power will gravitate toward the largest 34B model, but the 7B and 13B models can operate quickly on lower-powered hardware, and will handle tasks requiring low latency.
Meta is also serving up two variations on the tool, Code Llama – Python and Code Llama – Instruct. The former is specialized for Python code, while the latter has been tweaked to “generate helpful and safe answers in natural language.”
Anyone who wants a deeper breakdown of how Code Llama works (along with Meta’s approach to responsible A.I. use) can check out the company’s dedicated Code Llama page.
Meta has taken a relatively open approach to its A.I. development in recent months, with CEO Mark Zuckerberg pledging to make the fruits of the company’s research widely available to researchers, developers, and other tech pros. That stands in contrast to tech giants such as Google and Microsoft, which are more guarded about exposing their A.I.-related code to the public.
“When software is open, more people can scrutinize it to identify and fix potential issues,” Zuckerberg has written. Compare that to A.I. researchers who fear that releasing too much code too soon could result in A.I. being modified for nefarious means.
Meanwhile, employers are increasingly interested in integrating generative A.I. into their respective tech stacks. If you’re a tech professional who codes, boosting your knowledge of generative A.I. platforms and frameworks such as ChatGPT and TensorFlow can benefit your career (and potentially make your current workflow more efficient). While the number of A.I.-related jobs remains relatively small at the moment, expect that to change in coming years as more companies figure out how to use A.I. to their best advantage.
