tf.reduce_xxx函数

冷不防 2022-04-16 05:44 95阅读 0赞

\#书上称这一系列为规约函数。基本上都是降维

import tensorflow as tf
        import numpy as np
        t=np.random.randint(0,3,[3,4])
        sess=tf.Session()
        print("Test matrix is:\n",t)
        print("现在测试tf.reduce_sum，对tensor中的元素求和")
        print("tf.reduce_sum():",sess.run(tf.reduce_sum(t)))
        print("tf.reduce_sun(axis=0):",sess.run(tf.reduce_sum(t,axis=0)))
        print("tf.reduce_sun(axis=1):",sess.run(tf.reduce_sum(t,axis=1)))
        print("-----------------------------------------------------------------------------------------------")
        print("现在测试tf.reduce_prod，对tensor中的元素求乘积")
        print("tf.reduce_prod():",sess.run(tf.reduce_prod(t)))
        print("tf.reduce_prod(axis=0):",sess.run(tf.reduce_prod(t,axis=0)))
        print("tf.reduce_prod(axis=1):",sess.run(tf.reduce_prod(t,axis=1)))
        print("tf.reduce_prod(axis=0,keep_dims=True):",sess.run(tf.reduce_prod(t,axis=0,keep_dims=True)))
        print("tf.reduce_prod(axis=1,keep_dims=True):",sess.run(tf.reduce_prod(t,axis=1,keep_dims=True)))
        print("输出提示keep_dims将从以后的TF中移除，所以下面的测试不再测试这个参数，默认为False")
        print("-----------------------------------------------------------------------------------------------")
        print("现在测试tf.reduce_min,对tensor中的元素求最小值，reduce_max参数意义相同，忽略测试")
        print("tf.reduce_min():",sess.run(tf.reduce_min(t)))
        print("tf.reduce_min(axis=0):",sess.run(tf.reduce_min(t,axis=0)))
        print("tf.reduce_min(axis=1):",sess.run(tf.reduce_min(t,axis=1)))
        print("-----------------------------------------------------------------------------------------------")
        print("现在测试tf.reduce_mean，对tensor中的元素求均值，如果tensor元素是整数，则计算结果自动只取整数部分")
        print("tf.reduce_mean():",sess.run(tf.reduce_mean(t)))
        print("tf.reduce_mean(axis=0):",sess.run(tf.reduce_mean(t,axis=0)))
        print("tf.reduce_mean(axis=1):",sess.run(tf.reduce_mean(t,axis=1)))
        print("-----------------------------------------------------------------------------------------------")
        print("现在测试tf.reduce_all,对tensor中的元素求逻辑与")
        print("tf.reduce_all():",sess.run(tf.reduce_all(t)))
        print("tf.reduce_all(axis=0):",sess.run(tf.reduce_all(t,axis=0)))
        print("tf.reduce_all(axis=1):",sess.run(tf.reduce_all(t,axis=1)))
        print("-----------------------------------------------------------------------------------------------")
        print("现在测试tf.reduce_any,对tensor中的元素求逻辑或")
        print("tf.reduce_any():",sess.run(tf.reduce_any(t)))
        print("tf.reduce_any(axis=0):",sess.run(tf.reduce_any(t,axis=0)))
        print("tf.reduce_any(axis=1):",sess.run(tf.reduce_any(t,axis=1)))

[点击这里运行][Link 1]

[Link 1]: https://colab.research.google.com/drive/1R2p34uISUhpLBtI9ISZgYLhg2OH3Cl0o