当激活函数为softmax的时候，输出的值不应该是他们的和为1吗？代码地址
# 分类是否人脸的卷积输出层 conv4_1 = fluid.layers.conv2d(input=conv3_prelu, num_filters=2, filter_size=1, act='softmax', param_attr=ParamAttr(initializer=Xavier(), regularizer=L2DecayRegularizer(0.0005)), name='conv4_1')
# 分类是否人脸的卷积输出层 conv4_1 = fluid.layers.conv2d(input=conv3_prelu, num_filters=2, filter_size=1, act='softmax', param_attr=ParamAttr(initializer=Xavier(), regularizer=L2DecayRegularizer(0.0005)), name='conv4_1')
