:class:`IDTypeFeature` is a sparse matrix in `LIL <https://scipy-lectures.org/advanced/scipy_sparse/lil_matrix.html>`_ format which contains categ

r"""The :class:`PersiaBatch` is the type of dataset used internally in Persia. It wraps the :class:`IDTypeFeature`, :class:`NonIDTypeFeature`, :cl

Borrowing iterator over values of the linked array.

r"""Data context provides the communication functionality to data generator component. Used for sending a :class:`PersiaBatch` to the `nn worker`

:class:`Label` is the ``subclass`` of :class:`NdarrayDataBase` that you can add various datatype and multi-dimensional data. Example for :cla

r"""Streaming dataset receives the :class:`PersiaBatch` from the upstream data flow sent by :class:`DataCtx`. In the implemented :meth:`.Stre

r"""Initializes a common context for other persia context, e.g. :class:`DataCtx`, :class:`EmbeddingCtx` and :class:`TrainCtx`. This class should n

r"""Data loader will preprocess the data to the ``PersiaTrainingBatch``. The :class:`DataLoader` is a pipeline that preprocess the :class:`.Persi

The :class:`NdarrayDataBase` is a data structure that supports various datatypes and multi-dimensional data. PERSIA needs to convert the :class:`N

The :class:`NonIDTypeFeature` is the ``subclass`` of :class:`NdarrayDataBase` that you can add various datatypes and multi-dimensional data.

r"""A wrapper to config the embedding-server SGD optimizer.

r"""A wrapper to config the embedding-server Adagrad optimizer.

Implements an option to convert torch model to a DDP model. Current backend in :class:`DDPOption` only support `nccl` and `gloo`. You can set

r"""Embedding hyperparameters, argument of :class:`.EmbeddingCtx`.

r"""Provides the embedding-related functionality. :class:`EmbeddingCtx` can run offline test or online inference depending on different preprocess

The :class:`IDTypeFeatureWithSingleID` is a special format of :class:`IDTypeFeature` where there is only one id for each sample in the batch. :cla

r"""Subclass of :class:`EmbeddingCtx` that provides the inference functionality without `nats-servers`. Example for :class:`InferCtx`: .. co

Owning iterator over values of the linked array.

Launch the required processes to mock the distributed PERSIA environment. You can add the embedding config file path or a dict that contains the

r"""Subclass of :class:`EmbeddingCtx` that implements a `backward` function to update the embeddings. Example for :class:`TrainCtx`: ..

r"""Log level filter. Arguments: level (int): filter log level. Only logs with level higher than ``level`` will be kept.

r"""A wrapper to config the embedding-server Adam optimizer.

Implements an option to convert torch model to a bagua distributed model. Example for :class:`BaguaDistributedOption`: .. code-block:: pytho

Implements a common option to convert torch model to a distributed data parallel model, e.g. Bagua Distributed or pyTorch DDP. This class sho