This repository was archived by the owner on Nov 17, 2023. It is now read-only.
Replies: 1 comment
-
I shan't call this an answer until I know I can do batches, but this does at least work for the simple case without batches:
|
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I would like to use the linalg operators from C++, but they don't get exposed in op.h (i'm currently using 1.7 iirc)
That need not be a problem of course, because I would guess I can get at them through mxnet::Operator. I'm just a little unclear on exactly how I would do that however.
Specifically, I want to take a tensor (batch, 4, n) of vertices V and multiply by a transformation matrix (4,4) T, so V2 = V*M (and also, elsewhere, I want to do much the same, but with the inverse of another transformation matrix).
I'm fairly sure I'll be wanting to do something like:
mxnet::Operator("linalg_batch_gemm").SetInput(...).CreateSymbol()
I'm just hoping to get some help on the specifics of what to put in SetInput, and if there's any other quirks or "gotchas" I should be careful of.
Any help most appreciated, thanks.
Beta Was this translation helpful? Give feedback.
All reactions