You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Relay can run the SimplifyInference pass to simplify batch norm operators into a collection of primitive ops. Ideally, Glenside would intake these primitive ops and do efficient codegen for them. I don't think this is on the critical path for ISCA right now, and our workaround is to just take in opaque "BatchNormInference" nodes.
The text was updated successfully, but these errors were encountered:
We will be bringing batch norms into Glenside as these BatchNormInference nodes,
rather than running SimplifyInference in Relay and bringing them in as
adds/multiplies/divides/etc. This commit adds support for this new construct,
and parses it from Relay. I've added a flag to the Relay-to-Glenside compiler to
try and call this out specifically as a hack.
In the future, ideally, we'd bring in either batch norm nodes, or we'd bring in
the simplified components of batch norm inference and do efficient codegen for
them.
See #55
Relay can run the
SimplifyInference
pass to simplify batch norm operators into a collection of primitive ops. Ideally, Glenside would intake these primitive ops and do efficient codegen for them. I don't think this is on the critical path for ISCA right now, and our workaround is to just take in opaque "BatchNormInference" nodes.The text was updated successfully, but these errors were encountered: