Python API

Overview

This API section details functions, modules, and objects included in Apache MXNet, describing what they are and what they do. The APIs are grouped into the following categories:

Imperative API

mxnet.npnp/index.html

MXNet NP module, an array library that provides NumPy-compatible API.

mxnet.npxnpx/index.html

MXNet NPX module, neural network extension to the NumPy-compatible API.

mxnet.gluongluon/index.html

Imperative APIs to load data, construct and train neural networks.

Advanced modules

mxnet.runtimeruntime/index.html

API for querying MXNet enabled features.

mxnet.devicedevice/index.html

MXNet array device for specifying in-memory storage device.

mxnet.profilerprofiler/index.html

MXNet memory and performance profiler.

mxnet.executorexecutor/index.html

Managing symbolic graph execution.

mxnet.kvstore_serverkvstore_server/index.html

Server node for the key value store.

mxnet.engineengine/index.html

Engine properties management.

mxnet.rtcrtc/index.html

Tools for compiling and running CUDA code from the python frontend.

mxnet.test_utilstest_utils/index.html

Tools for using and testing MXNet.

mxnet.utilutil/index.html

General utility functions

Legacy

mxnet.ndarraylegacy/ndarray/index.html

Imperative APIs to manipulate multi-dimensional arrays.

mxnet.symbollegacy/symbol/index.html

Symbolic APIs for multi-dimensional arrays and neural network layers

mxnet.callbacklegacy/callback/index.html

Functions to track various statuses during an epoch.

mxnet.imagelegacy/image/index.html

Image iterators and image augmentation functions.

mxnet.iolegacy/io/index.html

Data iterators for common data formats and utility functions.

mxnet.recordiolegacy/recordio/index.html

Read and write for the RecordIO data format.

mxnet.visualizationlegacy/visualization/index.html

Functions for Symbol visualization.