Developer Guide for Intel® Data Analytics Acceleration Library 2019 Update 1
The forward reshape layer generates a tensor from input argument X of size n 1 x n 2 x ... x n p with modified size of dimensions m 1 x m 2 x ... x m q without modifying data and its order.
Given the dimension sizes of the output tensor m 1 x m 2 x ... x m q , such that
There are two reserved values of m i :
0 means m i = n i , valid only when i ≤ p
undefinedDimensionSize means m i is calculated to make the product of all mi equal to the product of all n i, valid only if used once. Corresponding m i calculated as follows:
The q-dimensional result tensor Y of size m 1 x ... x m q contains the same data from X in unmodified order.