API
=========================

.. toctree::
   :glob:
   :maxdepth: 1
   :hidden:

   mindspore
   mindspore.device_context
   mindspore.nn
   mindspore.ops
   mindspore.ops.primitive
   mindspore.mint
   mindspore.amp
   mindspore.train
   mindspore.communication
   mindspore.communication.comm_func
   mindspore.parallel
   mindspore.common.initializer
   mindspore.runtime
   mindspore.dataset
   mindspore.nn.probability
   mindspore.rewrite
   mindspore.multiprocessing
   mindspore.boost
   mindspore.numpy
   mindspore.scipy
   mindspore.utils
   mindspore.hal
   mindspore.experimental
   env_var_list
   ../note/api_mapping/pytorch_api_mapping

MindSpore provides rich interfaces for model building, training, and inference. The functions of each module interface are described below.

.. list-table::
   :widths: 40 60
   :header-rows: 1

   * - Module Name
     - Descriptions
   * - `mindspore <./mindspore.html>`_
     - Framework foundation interface.
   * - `mindspore.nn <./mindspore.nn.html>`_
     - A neural network layer for building predefined building blocks or computational units in a neural network.
   * - `mindspore.ops <./mindspore.ops.html>`_
     - Function interface.
   * - `mindspore.ops.primitive <./mindspore.ops.primitive.html>`_
     - Primitive operator.
   * - `mindspore.mint <./mindspore.mint.html>`_
     - Functional, nn, and optimizer interfaces consistent with mainstream industry usage.
   * - `mindspore.amp <./mindspore.amp.html>`_
     - Mixed-precision interface.
   * - `mindspore.train <./mindspore.train.html>`_
     - Traning interface.
   * - `mindspore.communication <./mindspore.communication.html>`_
     - Collection communication interface.
   * - `mindspore.communication.comm_func <./mindspore.communication.comm_func.html>`_
     - Collection communication functional interface.
   * - `mindspore.common.initializer <./mindspore.common.initializer.html>`_
     - Parameter initialization.
   * - `mindspore.hal <./mindspore.hal.html>`_
     - Interface of device management, stream management, event management and memory management.
   * - `mindspore.dataset.loading <./mindspore.dataset.loading.html>`_
     - Interfaces for loading and processing various datasets.
   * - `mindspore.dataset.transforms <./mindspore.dataset.transforms.html>`_
     - Generalized data transformations.
   * - `mindspore.mindrecord <./mindspore.mindrecord.html>`_
     - Operations interface related to efficient data format MindRecord developed by MindSpore.
   * - `mindspore.nn.probability <./mindspore.nn.probability.html>`_
     - Parameterizable probability distributions and sampling functions.
   * - `mindspore.rewrite <./mindspore.rewrite.html>`_
     - Custom rule-based model source code modification interface.
   * - `mindspore.multiprocessing <./mindspore.multiprocessing.html>`_
     - Multi-processing interface.
   * - `mindspore.boost <./mindspore.boost.html>`_
     - Automatic acceleration network interfaces.
   * - `mindspore.numpy <./mindspore.numpy.html>`_
     - NumPy Class interface.
   * - `mindspore.scipy <./mindspore.scipy.html>`_
     - SciPy Class interface.
   * - `mindspore.utils <./mindspore.utils.html>`_
     - Tools interface.
   * - `mindspore.experimental <./mindspore.experimental.html>`_
     - Experimental interface.
   * - `Environment Variables <./env_var_list.html>`_
     - Notes related to environment variables.