Dive into Deep Learning
Table Of Contents
Dive into Deep Learning
Table Of Contents

MXNet documentation

Due to the length of this book, it is impossible for us to introduce all MXNet functions and classes. The API documentation and additional tutorials and examples provide plenty of documentation beyond the book.

Finding all the functions and classes in the module

In order to know which functions and classes can be called in a module, we use the dir function. For instance we can query all the members or properties in the nd.random module.

In [1]:
from mxnet import nd
print(dir(nd.random))
['NDArray', '_Null', '__all__', '__builtins__', '__cached__', '__doc__', '__file__', '__loader__', '__name__', '__package__', '__spec__', '_internal', '_random_helper', 'current_context', 'exponential', 'gamma', 'generalized_negative_binomial', 'multinomial', 'negative_binomial', 'normal', 'numeric_types', 'poisson', 'shuffle', 'uniform']

Generally speaking, we can ignore functions that start and end with __ (special objects in Python) or functions that start with a single _(usually internal functions). According to the remaining member names, we can then hazard a guess that this module offers a generation method for various random numbers, including uniform distribution sampling (uniform), normal distribution sampling (normal), and Poisson sampling (poisson).

Finding the usage of specific functions and classes

For specific function or class usage, we can use the help function. Let’s take a look at the usage of the ones_like,  such as the NDArray function as an example.

In [2]:
help(nd.ones_like)
Help on function ones_like:

ones_like(data=None, out=None, name=None, **kwargs)
    Return an array of ones with the same shape and type
    as the input array.

    Examples::

      x = [[ 0.,  0.,  0.],
           [ 0.,  0.,  0.]]

      ones_like(x) = [[ 1.,  1.,  1.],
                      [ 1.,  1.,  1.]]



    Parameters
    ----------
    data : NDArray
        The input

    out : NDArray, optional
        The output NDArray to hold the result.

    Returns
    -------
    out : NDArray or list of NDArrays
        The output of this function.

From the documentation, we learned that the ones_like function creates a new one with the same shape as the NDArray and an element of 1. Let’s verify it:

In [3]:
x = nd.array([[0, 0, 0], [2, 2, 2]])
y = x.ones_like()
y
Out[3]:

[[1. 1. 1.]
 [1. 1. 1.]]
<NDArray 2x3 @cpu(0)>

In the Jupyter notebook, we can use ? to display the document in another window. For example, nd.ones_like? will create content that is almost identical to help(nd.ones_like), but will be displayed in an extra window. In addition, if we use two nd.ones_like??, the function implementation code will also be displayed.

API Documentation

For further details on the API details check the MXNet website at http://mxnet.apache.org/. You can find the details under the appropriate headings (also for programming languages other than Python).

Exercise

Check out the documentation for ones_like and for autograd.

Discuss on our Forum