1. 基本特點

1.1 save computation(惰性運行)

x = 2 y = 3 add_op = tf.add(x, y) mul_op = tf.multiply(x, y) useless = tf.multiply(x, add_op) pow_op = tf.pow(add_op, mul_op) with tf.Session() as sess:z = sess.run(pow_op)

如上，因為sess.run(pow_op)不需要用到useless，所以useless的運算不會被執行(save computation)

1.2 Distributed Computation

大圖分解為子圖多GPU并行計算, e.g.

freamework舉例

Multiple graphs有很多問題，如果非要實現Multiple graphs，可以考慮It’s better to have disconnected subgraphs within one graph

創建一個Graph

some codes,待整理

Why Graph:

Save computation. Only run subgraphs that lead to the values you want to fetch.

Break computation into small, differential pieces to facilitate auto-differentiation

Facilitate distributed computation, spread the work across multiple CPUs, GPUs, TPUs, or other devices

Many common machine learning models are taught and visualized as directed graphs

轉載于:https://www.cnblogs.com/LS1314/p/10366146.html

總結

以上是生活随笔為你收集整理的cs20_1-1的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。