Sometimes I just open the NumPy or the Python docs and browse it (or other libraries like PyTorch). Even after about 2 decades of Python experience, I often find something new and neat even on the "basic" pages that I didn't know about.
Reading release notes can be useful also.
Many people seem to have an irrational aversion to reading documentation and manuals. They'd rather speculate and guess and state false things than to just open the docs to understand how the thing works.
I'd say if you use the tool daily for years, it makes sense to invest into actually reading the docs.