-
Notifications
You must be signed in to change notification settings - Fork 63
Kanvi/Redo BatchMatMul #65
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
…anslating for other backends
…w/ngraph-bridge into kanvi/redo-batchmatmul
when number of dimensions = 3 and backend = CPU Add logic to handle the attributes correctly for the same Redo the tests for BatchMatMul Add more tests
…w/ngraph-bridge into kanvi/redo-batchmatmul
sayantan-nervana
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good in general. Lets wait for CPU's feedback from Gauri.
Minor comment about one of the tests
sindhu-nervana
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
sayantan-nervana
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
* Implement keep_dims for mean and sum; add a few tactical hacks for cases where DT_STRING is being placed on nGraph (not yet a complete solution) * Revert DT_STRING stuff since #68 fixes the issue * Remove tf_keep_dims 'Unimplemented' check for Mean in builder
* Redo BatchMatMul since nGraph has the op supposrt for it now and add tests * Skip BatchMatMul3* tests for GPU since GPU does not support this op * Changes to support nGraph BatchMatMul for CPU and the older way of translating for other backends * Redo the translation for BatchMatMul so that it uses the CPU BatchMatMul when number of dimensions = 3 and backend = CPU Add logic to handle the attributes correctly for the same Redo the tests for BatchMatMul * Reorganize the code to add the correct backend as attribute * Add logic to pass higher-order(>3) BatchMatMul to CPU as well. * Update test/test_math_ops.cpp
Redo BatchMatMul since nGraph has the op support for it now
Add tests