8e4926a58d0ff919db9cb85ac9530053eda62190,models/VGGBinomialDropout.py,VGGBinomialDropout,_inference,#VGGBinomialDropout#Any#Any#Any#Any#Any#,20

Before Change

        with tf.variable_scope(self.__class__.__name__):
            with tf.variable_scope("64"):
                with tf.variable_scope("conv1"):
                    conv1 = tf.nn.relu(
                        utils.conv_layer(
                            images, [3, 3, 3, 64], 1, "SAME", wd=l2_penalty))
                    if train_phase:
                        conv1 = utils.binomial_dropout(conv1, 0.7)

                with tf.variable_scope("conv2"):
                    conv2 = tf.nn.relu(
                        utils.conv_layer(
                            conv1, [3, 3, 64, 64], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv2 = utils.binomial_dropout(conv2, 0.6)

            with tf.variable_scope("pool1"):
                pool1 = tf.nn.max_pool(
                    conv2,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("128"):
                with tf.variable_scope("conv3"):
                    conv3 = tf.nn.relu(
                        utils.conv_layer(
                            pool1, [3, 3, 64, 128], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv3 = utils.binomial_dropout(conv3, 0.6)

                with tf.variable_scope("conv4"):
                    conv4 = tf.nn.relu(
                        utils.conv_layer(
                            conv3, [3, 3, 128, 128], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv4 = utils.binomial_dropout(conv4, 0.6)

            with tf.variable_scope("pool2"):
                pool2 = tf.nn.max_pool(
                    conv4,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("256"):
                with tf.variable_scope("conv5"):
                    conv5 = tf.nn.relu(
                        utils.conv_layer(
                            pool2, [3, 3, 128, 256], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv5 = utils.binomial_dropout(conv5, 0.6)

                with tf.variable_scope("conv6"):
                    conv6 = tf.nn.relu(
                        utils.conv_layer(
                            conv5, [3, 3, 256, 256], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv6 = utils.binomial_dropout(conv6, 0.6)

                with tf.variable_scope("conv7"):
                    conv7 = tf.nn.relu(
                        utils.conv_layer(
                            conv6, [3, 3, 256, 256], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv7 = utils.binomial_dropout(conv7, 0.6)

            with tf.variable_scope("pool3"):
                pool3 = tf.nn.max_pool(
                    conv7,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("512"):
                with tf.variable_scope("conv8"):
                    conv8 = tf.nn.relu(
                        utils.conv_layer(
                            pool3, [3, 3, 256, 512], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv8 = utils.binomial_dropout(conv8, 0.6)

                with tf.variable_scope("conv9"):
                    conv9 = tf.nn.relu(
                        utils.conv_layer(
                            conv8, [3, 3, 512, 512], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv9 = utils.binomial_dropout(conv9, 0.6)

                with tf.variable_scope("conv10"):
                    conv10 = tf.nn.relu(
                        utils.conv_layer(
                            conv9, [3, 3, 512, 512], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv10 = utils.binomial_dropout(conv10, 0.6)

            with tf.variable_scope("pool4"):
                pool4 = tf.nn.max_pool(
                    conv10,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("512b2"):
                with tf.variable_scope("conv11"):
                    conv11 = tf.nn.relu(
                        utils.conv_layer(
                            pool4, [3, 3, 512, 512], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv11 = utils.binomial_dropout(conv11, 0.6)

                with tf.variable_scope("conv12"):
                    conv12 = tf.nn.relu(
                        utils.conv_layer(
                            conv11, [3, 3, 512, 512], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv12 = utils.binomial_dropout(conv12, 0.6)

                with tf.variable_scope("conv13"):
                    conv13 = tf.nn.relu(
                        utils.conv_layer(
                            conv12, [3, 3, 512, 512], 1, "SAME", wd=l2_penalty))

                    if train_phase:
                        conv13 = utils.binomial_dropout(conv13, 0.6)

            with tf.variable_scope("pool5"):
                pool5 = tf.nn.max_pool(
                    conv13,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

                pool5 = tf.reshape(pool5, [-1, 512])

            with tf.variable_scope("fc"):
                fc1 = tf.nn.relu(
                    utils.fc_layer(
                        pool5, [512, 512], wd=l2_penalty))

                if train_phase:
                    fc1 = utils.binomial_dropout(fc1, 0.5)

After Change

        with tf.variable_scope(self.__class__.__name__):
            with tf.variable_scope("64"):
                with tf.variable_scope("conv1"):
                    conv1 = utils.conv_layer(
                        images, [3, 3, 3, 64],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)
                    if train_phase:
                        conv1 = utils.binomial_dropout(conv1, 0.7)

                with tf.variable_scope("conv2"):
                    conv2 = utils.conv_layer(
                        conv1, [3, 3, 64, 64],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv2 = utils.binomial_dropout(conv2, 0.6)

            with tf.variable_scope("pool1"):
                pool1 = tf.nn.max_pool(
                    conv2,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("128"):
                with tf.variable_scope("conv3"):
                    conv3 = utils.conv_layer(
                        pool1, [3, 3, 64, 128],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv3 = utils.binomial_dropout(conv3, 0.6)

                with tf.variable_scope("conv4"):
                    conv4 = utils.conv_layer(
                        conv3, [3, 3, 128, 128],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv4 = utils.binomial_dropout(conv4, 0.6)

            with tf.variable_scope("pool2"):
                pool2 = tf.nn.max_pool(
                    conv4,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("256"):
                with tf.variable_scope("conv5"):
                    conv5 = utils.conv_layer(
                        pool2, [3, 3, 128, 256],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv5 = utils.binomial_dropout(conv5, 0.6)

                with tf.variable_scope("conv6"):
                    conv6 = utils.conv_layer(
                        conv5, [3, 3, 256, 256],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv6 = utils.binomial_dropout(conv6, 0.6)

                with tf.variable_scope("conv7"):
                    conv7 = utils.conv_layer(
                        conv6, [3, 3, 256, 256],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv7 = utils.binomial_dropout(conv7, 0.6)

            with tf.variable_scope("pool3"):
                pool3 = tf.nn.max_pool(
                    conv7,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("512"):
                with tf.variable_scope("conv8"):
                    conv8 = utils.conv_layer(
                        pool3, [3, 3, 256, 512],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv8 = utils.binomial_dropout(conv8, 0.6)

                with tf.variable_scope("conv9"):
                    conv9 = utils.conv_layer(
                        conv8, [3, 3, 512, 512],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv9 = utils.binomial_dropout(conv9, 0.6)

                with tf.variable_scope("conv10"):
                    conv10 = utils.conv_layer(
                        conv9, [3, 3, 512, 512],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv10 = utils.binomial_dropout(conv10, 0.6)

            with tf.variable_scope("pool4"):
                pool4 = tf.nn.max_pool(
                    conv10,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

            with tf.variable_scope("512b2"):
                with tf.variable_scope("conv11"):
                    conv11 = utils.conv_layer(
                        pool4, [3, 3, 512, 512],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv11 = utils.binomial_dropout(conv11, 0.6)

                with tf.variable_scope("conv12"):
                    conv12 = utils.conv_layer(
                        conv11, [3, 3, 512, 512],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv12 = utils.binomial_dropout(conv12, 0.6)

                with tf.variable_scope("conv13"):
                    conv13 = utils.conv_layer(
                        conv12, [3, 3, 512, 512],
                        1,
                        "SAME",
                        activation=tf.nn.relu,
                        wd=l2_penalty)

                    if train_phase:
                        conv13 = utils.binomial_dropout(conv13, 0.6)

            with tf.variable_scope("pool5"):
                pool5 = tf.nn.max_pool(
                    conv13,
                    ksize=[1, 2, 2, 1],
                    strides=[1, 2, 2, 1],
                    padding="VALID")

                pool5 = tf.reshape(pool5, [-1, 512])

            with tf.variable_scope("fc"):
                fc1 = utils.fc_layer(
                    pool5, [512, 512], activation=tf.nn.relu, wd=l2_penalty)

                if train_phase:
                    fc1 = utils.binomial_dropout(fc1, 0.5)

Instances