How good are LLMs at reducing code? For example, will they recognize a common problem and build an abstraction around it? I imagine that the solutions they produce tend to have a lot of repetition with small differences that could be improved by abstraction.
No comments yet.