How to use the babi.base_model.BaseTower function in babi

L = tf.tile(tf.expand_dims(tf.expand_dims(L, 0), 0), [N, d, 1, 1])
            sL = tf.tile(tf.expand_dims(tf.expand_dims(sL, 0), 0), [N, d, 1, 1])
            logb = tf.log(b + 1e-9)  # [N, M, d]
            logb = tf.concat(1, [tf.zeros([N, 1, d]), tf.slice(logb, [0, 1, 0], [-1, -1, -1])])  # [N, M, d]
            logb = tf.expand_dims(tf.transpose(logb, [0, 2, 1]), -1)  # [N, d, M, 1]
            left = L * tf.exp(tf.batch_matmul(L, logb * sL))  # [N, d, M, M]
            right = a * u_t  # [N, M, d]
            right = tf.expand_dims(tf.transpose(right, [0, 2, 1]), -1)  # [N, d, M, 1]
            u = tf.batch_matmul(left, right)  # [N, d, M, 1]
            u = tf.transpose(tf.squeeze(u, [3]), [0, 2, 1])  # [N, M, d]
            print ("L : %s\nsL: %s,\nlogb : %s\na : %s, b : %s, u_t : %s, left : %s, right : %s" % (L,sL, logb, a, b, u_t, left, right))
            assert False
        return u


class Tower(BaseTower):
    def initialize(self):
        params = self.params
        placeholders = self.placeholders
        tensors = self.tensors
        variables_dict = self.variables_dict
        N, J, V, Q, M = params.batch_size, params.max_sent_size, params.vocab_size, params.max_ques_size, params.mem_size
        d = params.hidden_size
        L = params.mem_num_layers
        att_forget_bias = params.att_forget_bias
        use_vector_gate = params.use_vector_gate
        wd = params.wd
        initializer = tf.random_uniform_initializer(-np.sqrt(3), np.sqrt(3))
        with tf.name_scope("placeholders"):
            x = tf.placeholder('int32', shape=[N, M, J], name='x')
            x_mask = tf.placeholder('bool', shape=[N, M, J], name='x_mask')
            q = tf.placeholder('int32', shape=[N, J], name='q')