A config builder for a input or a weight.

InputSpec describes the signature information of the model input, such as ``shape`` , ``dtype`` , ``name`` . This interface is often used to

A config builder for generating a Op.

Create Python Program. It has at least one :ref:`api_guide_Block_en`, when the control flow op like conditional_block, while :ref:`api_paddl

A config builder for generating a Program. input_type : (np.dtype, default=None), the inputs will be casted to input_type before f

r""" Fully-connected linear transformation layer. For each input :math:`X` , the equation is: .. math:: Out = XW + b where

* A unified data class for inferring distributed attributes * in both dygraph mode and static mode */

DummyTracker is a subclass of Tracker that specifically tracks variables cannot be reproduced from the frame. It is mostly generated by compl

Note: ``gradient_clip`` of ``ParamAttr`` HAS BEEN DEPRECATED since 2.0. Please use ``need_clip`` in ``ParamAttr`` to specify the

DataLoader provides an iterator which iterates given dataset once by the batch_sampler. DataLoader supports single-process and multi-pro

DanglingTracker is a subclass of Tracker that specifically tracks variables that are not in the frame. Variables whose tracker is DanglingTra

ConstantVariable is a subclass of VariableBase used to wrap a Variable of the const type. Args: value(Any): The value to be wrapped.

A named variable. i ∈ [lower_bound, upper_bound)

BuiltinVariable is a subclass of FunctionVariable used to wrap a built-in function. Args: fn (Callable[..., Any]): The built-in funct

Represents a local shard of a distributed tensor parameter. Args: key (str): The name of the parameter. local_tensor (Tensor

A set that preserves the order of insertion.

docstring for Conv2D

Configurable block used for Convolution-Normalization-Activation blocks. This code is based on the torchvision code with modifications. Y

Implement the constant initializer. Args: value (float32|float64, optional): constant value to initialize the parameter. Default: 0.0.

:api_attr: Static Graph An Executor in Python, supports single/multiple-GPU running, and single/multiple-CPU running. Args:

device-agnostic abstraction of any-stream

GetAttrTracker is a subclass of Tracker that specifically tracks the attribute access of an variable. Args: obj (VariableBase): The

* A struct specifying a collection of outputs. */

r""" The class `ProcessMesh` describes the topology of logical processes. Args: mesh (list|numpy.array): an N-dimensional array descr

* Allocate a buffer with the given type and size. The buffer lives for at most * the duration of the body statement, within which it is freed. */

* Types in the CINN type system. They can be ints, unsigned ints, or floats of * various bit-widths. They can also be vectors of the same (by setting

* The sum of two expressions. */

A Model object is a network with training and inference features. Dynamic graph and static graph are supported at the same time, switche

r""" This interface is used to construct a callable object of the ``Conv2D`` class. For more details, refer to code examples. The convolut