anywords.py 文件源码

def compile_model(self):
        self.model.compile(loss=losses.categorical_crossentropy,
                           optimizer=optimizers.rmsprop(),
                           metrics=['accuracy'])
        self.graph = tf.get_default_graph()