當前位置: 首頁>>代碼示例>>Python>>正文

Python ndarray.pad方法代碼示例

本文整理匯總了Python中mxnet.ndarray.pad方法的典型用法代碼示例。如果您正苦於以下問題：Python ndarray.pad方法的具體用法？Python ndarray.pad怎麽用？Python ndarray.pad使用的例子？那麽, 這裏精選的方法代碼示例或許可以為您提供幫助。您也可以進一步了解該方法所在類mxnet.ndarray的用法示例。


在下文中一共展示了ndarray.pad方法的5個代碼示例，這些例子默認根據受歡迎程度排序。您可以為喜歡或者感覺有用的代碼點讚，您的評價將有助於係統推薦出更棒的Python代碼示例。

示例1: forward

# 需要導入模塊: from mxnet import ndarray [as 別名]
# 或者: from mxnet.ndarray import pad [as 別名]
def forward(self, x):
        return F.pad(x, mode='reflect', pad_width=self.pad_width)
開發者ID:awslabs，項目名稱:dynamic-training-with-apache-mxnet-on-aws，代碼行數:4，代碼來源:net.py

示例2: __init__

# 需要導入模塊: from mxnet import ndarray [as 別名]
# 或者: from mxnet.ndarray import pad [as 別名]
def __init__(self, in_channels, out_channels, kernel_size, stride):
        super(ConvLayer, self).__init__()
        padding = int(np.floor(kernel_size / 2))
        self.pad = ReflectancePadding(pad_width=(0,0,0,0,padding,padding,padding,padding))
        self.conv2d = nn.Conv2D(in_channels=in_channels, channels=out_channels, 
                                kernel_size=kernel_size, strides=(stride,stride),
                                padding=0)
開發者ID:awslabs，項目名稱:dynamic-training-with-apache-mxnet-on-aws，代碼行數:9，代碼來源:net.py

示例3: hybrid_forward

# 需要導入模塊: from mxnet import ndarray [as 別名]
# 或者: from mxnet.ndarray import pad [as 別名]
def hybrid_forward(self, F, x):
        x = x.transpose(axes=(0, 2, 1, 3))
        x = F.pad(x, mode="constant", pad_width=(0, 0, 0, 0, 0, self.padding, 0, 0), constant_value=0)
        x = x.transpose(axes=(0, 2, 1, 3))
        return x
開發者ID:osmr，項目名稱:imgclsmob，代碼行數:7，代碼來源:irevnet.py

示例4: __init__

# 需要導入模塊: from mxnet import ndarray [as 別名]
# 或者: from mxnet.ndarray import pad [as 別名]
def __init__(self,
                 in_channels,
                 out_channels,
                 strides,
                 bn_use_global_stats,
                 preactivate,
                 **kwargs):
        super(IRevUnit, self).__init__(**kwargs)
        if not preactivate:
            in_channels = in_channels // 2

        padding = 2 * (out_channels - in_channels)
        self.do_padding = (padding != 0) and (strides == 1)
        self.do_downscale = (strides != 1)

        with self.name_scope():
            if self.do_padding:
                self.pad = IRevInjectivePad(padding)
            self.bottleneck = IRevBottleneck(
                in_channels=in_channels,
                out_channels=out_channels,
                strides=strides,
                bn_use_global_stats=bn_use_global_stats,
                preactivate=preactivate)
            if self.do_downscale:
                self.psi = IRevDownscale(strides)
開發者ID:osmr，項目名稱:imgclsmob，代碼行數:28，代碼來源:irevnet.py

示例5: inverse

# 需要導入模塊: from mxnet import ndarray [as 別名]
# 或者: from mxnet.ndarray import pad [as 別名]
def inverse(self, x2, y1):
        import mxnet.ndarray as F

        if self.do_downscale:
            x2 = self.psi.inverse(x2)
        fx2 = - self.bottleneck(x2)
        x1 = fx2 + y1
        if self.do_downscale:
            x1 = self.psi.inverse(x1)
        if self.do_padding:
            x = F.concat(x1, x2, dim=1)
            x = self.pad.inverse(x)
            x1, x2 = F.split(x, axis=1, num_outputs=2)
        return x1, x2
開發者ID:osmr，項目名稱:imgclsmob，代碼行數:16，代碼來源:irevnet.py


注：本文中的mxnet.ndarray.pad方法示例由純淨天空整理自Github/MSDocs等開源代碼及文檔管理平台，相關代碼片段篩選自各路編程大神貢獻的開源項目，源碼版權歸原作者所有，傳播和使用請參考對應項目的License；未經允許，請勿轉載。