All possible signatures with numba vectorize

Accelerated Computing CUDA CUDA Programming and Performance
1

I am new to Numba and GPU programming with Python.

I understand basic signatures used with vectorize decorator.
For example :- For a method that accepts two float variables and return a float , this is the signature

@vectorize([float64(float64, float64)])

I have 3 questions
1.What is the signature to accept and return Numpy arrays?
2.What is the signature to accept and return Pandas Dataframe?
3.What are the possible signatures available ?