How to implement tf.clip_by_value() when converted the TF PB model to DLC model

如何在SNPE中实现Tensorfow的tf.clip_by_value()操作

将下面的代码替换掉tf.clip_by_value()，即可实现tf pb模型的转dlc模型。
You can use the above fuction clip_by_value to replace the **tf.clip_by_value()**function. Then it will work when you convert the tf pb model to SNPE DLC model.

def clip_by_value(x, min, max, name='clip'):
    with tf.name_scope(name=name):
        with tf.name_scope(name='clipmin'):
            res = tf.maximum(x, min)
            res = -1. * res
        with tf.name_scope(name='clipmax'):
            max = -1. * max
            res = tf.maximum(res, max)
        with tf.name_scope(name='clip_res'):
            res = -1. * res
        return res