experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update documentation for basic Dot operation

Browse Source 
The current documentation does not match the actual implementation in XlaBuilder

The implementation sets lhs_contracting_dimensions to 0 if the rank of lhs is 1, otherwise 1. It always sets rhs_contracting_dimensions to 0.

PiperOrigin-RevId: 244371170
This commit is contained in:
A. Unique TensorFlower 2019-04-19 09:36:16 -07:00 committed by TensorFlower Gardener
parent 653dfe126c
commit b6e4bf24eb
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
tensorflow/compiler/xla/g3doc/operation_semantics.md
View File

@ -928,11 +928,11 @@ The exact semantics of this operation depend on the ranks of the operands:
| matrix [m x k] `dot` | matrix [m x n] | matrix-matrix | | matrix [m x k] `dot` | matrix [m x n] | matrix-matrix |
: matrix [k x n] : : multiplication : : matrix [k x n] : : multiplication :
The operation performs sum of products over the last dimension of `lhs` and the The operation performs sum of products over the second dimension of `lhs` (or
one-before-last dimension of `rhs`. These are the "contracted" dimensions. The the first if it has rank 1) and the first dimension of `rhs`. These are the
contracted dimensions of `lhs` and `rhs` must be of the same size. In practice, "contracted" dimensions. The contracted dimensions of `lhs` and `rhs` must be of
it can be used to perform dot products between vectors, vector/matrix the same size. In practice, it can be used to perform dot products between
multiplications or matrix/matrix multiplications. vectors, vector/matrix multiplications or matrix/matrix multiplications.
## DotGeneral ## DotGeneral