Tensorflow中的tf.matmul和tf.batch_matmul有什么区别？

Question

For example if I have the following data:

x = tf.placeholder("float", [None, n, n])
y = tf.placeholder("float", [None, n, n])

Is there any difference between the two operations?

res = tf.matmul(x,y)
res = tf.batch_matmul(x,y)

Answer 1

tf.batch_matmul is deprecated in favor of tf.matmul in version 0.12 and later so no difference in later versions. Earlier versions required rank-2 inputs for matmul but allowed larger ranks for batch_matmul