Researchers at OpenAI are evaluating large language models that have been trained on code. These models can perform tasks such as writing code and explaining it, but their accuracy and reliability vary widely depending on the specific model and task. The evaluation process involves testing these models on a range of programming languages and assessing their ability to write correct and readable code. --- Why it matters: This matters because large language models trained on code have the potential to revolutionize software development by automating tasks such as coding and debugging, but their accuracy and reliability must be carefully evaluated before they can be trusted. Source: https://openai.com/index/evaluating-large-language-models-trained-on-code