.. _cn_api_fluid_BuildStrategy: BuildStrategy ------------------------------- :api_attr: 声明式编程模式(静态图) .. py:class:: paddle.fluid.BuildStrategy ``BuildStrategy`` 使用户更方便地控制 :ref:`cn_api_fluid_ParallelExecutor` 中计算图的建造方法,可通过设置 ``ParallelExecutor`` 中的 ``BuildStrategy`` 成员来实现此功能。 **代码示例** .. code-block:: python import os import numpy as np import paddle.fluid as fluid os.environ["CPU_NUM"] = '2' places = fluid.cpu_places() data = fluid.layers.data(name="x", shape=[1], dtype="float32") hidden = fluid.layers.fc(input=data, size=10) loss = fluid.layers.mean(hidden) fluid.optimizer.SGD(learning_rate=0.01).minimize(loss) build_strategy = fluid.BuildStrategy() build_strategy.enable_inplace = True build_strategy.memory_optimize = True build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce program = fluid.compiler.CompiledProgram(fluid.default_main_program()) program = program.with_data_parallel(loss_name=loss.name, build_strategy=build_strategy, places=places) .. py:attribute:: debug_graphviz_path str类型。表示以graphviz格式向文件中写入计算图的路径,有利于调试。默认值为空字符串。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.debug_graphviz_path = "./graph" .. py:attribute:: enable_sequential_execution bool类型。如果设置为True,则算子的执行顺序将与算子定义的执行顺序相同。默认为False。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.enable_sequential_execution = True .. py:attribute:: fuse_broadcast_ops bool类型。表明是否融合(fuse) broadcast ops。该选项指在Reduce模式下有效,使程序运行更快。默认为False。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.fuse_broadcast_ops = True .. py:attribute:: fuse_elewise_add_act_ops bool类型。表明是否融合(fuse) elementwise_add_op和activation_op。这会使整体执行过程更快。默认为False。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.fuse_elewise_add_act_ops = True .. py:attribute:: fuse_relu_depthwise_conv bool类型。表明是否融合(fuse) relu和depthwise_conv2d,节省GPU内存并可能加速执行过程。此选项仅适用于GPU设备。默认为False。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.fuse_relu_depthwise_conv = True .. py:attribute:: gradient_scale_strategy ``fluid.BuildStrategy.GradientScaleStrategy`` 类型。在 ``ParallelExecutor`` 中,存在三种定义loss对应梯度( *loss@grad* )的方式,分别为 ``CoeffNumDevice``, ``One`` 与 ``Customized``。默认情况下, ``ParallelExecutor`` 根据设备数目来设置 *loss@grad* 。如果用户需要自定义 *loss@grad* ,可以选择 ``Customized`` 方法。默认为 ``CoeffNumDevice`` 。 **代码示例** .. code-block:: python import os import numpy as np import paddle.fluid as fluid import paddle.fluid.compiler as compiler use_cuda = True place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace() exe = fluid.Executor(place) # NOTE: 如果你使用CPU计算,需要指定CPU_NUM, 否则,fluid # 将使用所有的核的数目作为CPU_NUM, # 这种情况下,输入的batch size应该大于CPU_NUM, 否则, # 进程将会因为异常而失败。 if not use_cuda: os.environ['CPU_NUM'] = str(2) places = fluid.cpu_places() else: places = places = fluid.cuda_places() data = fluid.layers.data(name='X', shape=[1], dtype='float32') hidden = fluid.layers.fc(input=data, size=10) loss = fluid.layers.mean(hidden) fluid.optimizer.SGD(learning_rate=0.01).minimize(loss) fluid.default_startup_program().random_seed=1 exe.run(fluid.default_startup_program()) build_strategy = fluid.BuildStrategy() build_strategy.gradient_scale_strategy = \ fluid.BuildStrategy.GradientScaleStrategy.Customized compiled_prog = compiler.CompiledProgram( fluid.default_main_program()).with_data_parallel( loss_name=loss.name, build_strategy=build_strategy, places = places) dev_count = len(places) x = np.random.random(size=(10, 1)).astype('float32') loss_grad = np.ones((dev_count)).astype("float32") * 0.01 loss_grad_name = loss.name+"@GRAD" loss_data = exe.run(compiled_prog, feed={"X": x, loss_grad_name : loss_grad}, fetch_list=[loss.name, loss_grad_name]) .. py:attribute:: memory_optimize bool类型或None。设为True时可用于减少总内存消耗,False表示不使用,None表示框架会自动选择使用或者不使用优化策略。当前,None意味着当GC不能使用时,优化策略将被使用。默认为None。 .. py:attribute:: reduce_strategy ``fluid.BuildStrategy.ReduceStrategy`` 类型。在 ``ParallelExecutor`` 中,存在两种参数梯度聚合策略,即 ``AllReduce`` 和 ``Reduce`` 。如果用户需要在所有执行设备上独立地进行参数更新,可以使用 ``AllReduce`` 。如果使用 ``Reduce`` 策略,所有参数的优化将均匀地分配给不同的执行设备,随之将优化后的参数广播给其他执行设备。 默认值为 ``AllReduce`` 。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce .. py:attribute:: remove_unnecessary_lock bool类型。设置True会去除GPU操作中的一些锁操作, ``ParallelExecutor`` 将运行得更快,默认为True。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.remove_unnecessary_lock = True .. py:attribute:: sync_batch_norm bool类型。表示是否使用同步的批正则化,即在训练阶段通过多个设备同步均值和方差。当前的实现不支持FP16训练和CPU。并且目前**仅支持**仅在一台机器上进行同步式批正则。默认为 False。 **代码示例** .. code-block:: python import paddle.fluid as fluid build_strategy = fluid.BuildStrategy() build_strategy.sync_batch_norm = True