de14a161eb265f546e22dfec99e0097e6dc0a7a2,python/mxnet_benchmarks/models/googlenet.py,GoogleNet,__init__,#GoogleNet#Any#,50

Before Change

        Model.__init__(self, params)
        training = self.phase == "training"

        data = mx.sym.Variable("data")

        conv1 = ConvFactory(data, 64, kernel=(7, 7), stride=(2, 2), pad=(3, 3), name="conv1/7x7_s2")
        pool1 = mx.sym.Pooling(conv1, kernel=(3, 3), stride=(2, 2), pool_type="max", name="pool1/3x3_s2")
        norm1 = mx.symbol.LRN(data=pool1, alpha=0.0001, beta=0.75, knorm=2, nsize=5, name="pool1/norm1")

        conv2_reduce = ConvFactory(norm1, 64, kernel=(1, 1), stride=(1, 1), name="conv2/3x3_reduce")

        conv2 = ConvFactory(conv2_reduce, 192, kernel=(3, 3), stride=(1, 1), pad=(1, 1), name="conv2/3x3")
        norm2 = mx.symbol.LRN(data=conv2, alpha=0.0001, beta=0.75, knorm=2, nsize=5, name="conv2/norm2")
        pool2 = mx.sym.Pooling(norm2, kernel=(3, 3), stride=(2, 2), pool_type="max", name="pool2/3x3_s2")

        in3a = InceptionFactory(pool2, 64, 96, 128, 16, 32, "max", 32, name="inception_3a")
        in3b = InceptionFactory(in3a, 128, 128, 192, 32, 96, "max", 64, name="inception_3b")
        pool3 = mx.sym.Pooling(in3b, kernel=(3, 3), stride=(2, 2), pool_type="max", name="pool3/3x3_s2")

        in4a = InceptionFactory(pool3, 192, 96, 208, 16, 48, "max", 64, name="inception_4a")
        in4b = InceptionFactory(in4a, 160, 112, 224, 24, 64, "max", 64, name="inception_4b")
        in4c = InceptionFactory(in4b, 128, 128, 256, 24, 64, "max", 64, name="inception_4c")
        in4d = InceptionFactory(in4c, 112, 144, 288, 32, 64, "max", 64, name="inception_4d")
        in4e = InceptionFactory(in4d, 256, 160, 320, 32, 128, "max", 128, name="inception_4e")
        pool4 = mx.sym.Pooling(in4e, kernel=(3, 3), stride=(2, 2), pad=(1, 1), pool_type="max", name="pool4/3x3_s2")

        in5a = InceptionFactory(pool4, 256, 160, 320, 32, 128, "max", 128, name="inception_5a")
        in5b = InceptionFactory(in5a, 384, 192, 384, 48, 128, "max", 128, name="inception_5b")
        pool5 = mx.sym.Pooling(in5b, kernel=(7, 7), stride=(1, 1), pool_type="avg", name="pool5/7x7_s1")
        flatten5 = mx.sym.Flatten(data=pool5)
        drop5 = mx.symbol.Dropout(data=flatten5, p=0.5, name="pool5/drop_7x7_s1") if training else flatten5

        self.__output = self.add_head_nodes(drop5)

After Change

        Model.__init__(self, params)
        training = self.phase == "training"

        data = self.add_data_node()

        conv1 = ConvFactory(data, 64, kernel=(7, 7), stride=(2, 2), pad=(3, 3), name="conv1/7x7_s2")
        pool1 = mx.sym.Pooling(conv1, kernel=(3, 3), stride=(2, 2), pool_type="max", name="pool1/3x3_s2")
        norm1 = mx.symbol.LRN(data=pool1, alpha=0.0001, beta=0.75, knorm=2, nsize=5, name="pool1/norm1")

        conv2_reduce = ConvFactory(norm1, 64, kernel=(1, 1), stride=(1, 1), name="conv2/3x3_reduce")

        conv2 = ConvFactory(conv2_reduce, 192, kernel=(3, 3), stride=(1, 1), pad=(1, 1), name="conv2/3x3")
        norm2 = mx.symbol.LRN(data=conv2, alpha=0.0001, beta=0.75, knorm=2, nsize=5, name="conv2/norm2")
        pool2 = mx.sym.Pooling(norm2, kernel=(3, 3), stride=(2, 2), pool_type="max", name="pool2/3x3_s2")

        in3a = InceptionFactory(pool2, 64, 96, 128, 16, 32, "max", 32, name="inception_3a")
        in3b = InceptionFactory(in3a, 128, 128, 192, 32, 96, "max", 64, name="inception_3b")
        pool3 = mx.sym.Pooling(in3b, kernel=(3, 3), stride=(2, 2), pool_type="max", name="pool3/3x3_s2")

        in4a = InceptionFactory(pool3, 192, 96, 208, 16, 48, "max", 64, name="inception_4a")
        in4b = InceptionFactory(in4a, 160, 112, 224, 24, 64, "max", 64, name="inception_4b")
        in4c = InceptionFactory(in4b, 128, 128, 256, 24, 64, "max", 64, name="inception_4c")
        in4d = InceptionFactory(in4c, 112, 144, 288, 32, 64, "max", 64, name="inception_4d")
        in4e = InceptionFactory(in4d, 256, 160, 320, 32, 128, "max", 128, name="inception_4e")
        pool4 = mx.sym.Pooling(in4e, kernel=(3, 3), stride=(2, 2), pad=(1, 1), pool_type="max", name="pool4/3x3_s2")

        in5a = InceptionFactory(pool4, 256, 160, 320, 32, 128, "max", 128, name="inception_5a")
        in5b = InceptionFactory(in5a, 384, 192, 384, 48, 128, "max", 128, name="inception_5b")
        pool5 = mx.sym.Pooling(in5b, kernel=(7, 7), stride=(1, 1), pool_type="avg", name="pool5/7x7_s1")
        flatten5 = mx.sym.Flatten(data=pool5)
        drop5 = mx.symbol.Dropout(data=flatten5, p=0.5, name="pool5/drop_7x7_s1") if training else flatten5

        self.__output = self.add_head_nodes(drop5)

Instances