Researchers open-source benchmarks measuring quality of AI-generated code

A new study finds that large AI language models are beginning to learn to code, albeit with a few mistakes.
https://bit.ly/2Tw3z6s