Python 3 vs Python 2 is easily answered: The differences are subtle (there are concise lists of differences), you can apply any old tutorial/book/Ressource and for instance mostly have to ensure to type print(foo) instead of print foo or foo.items() instead of foo.iteritems().
Depending on libraries, it highly depends on what you want to do. Python for data science? There are tons of great open books.
Python is great for its REPL shell. You can learn the language by experimenting and exploring. This approach is also part of many API docs in python, in contrast to many other comparable programming languages.