masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-579436246
do we still need partitioning if we get rid of the annotation pass? At least
for simple patterns like conv + bias + relu, I can get the
masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-579436246
do we still need partitioning if we get rid of the annotation pass? At least
for simple patterns like conv + bias + relu, I can get the
masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-578329595
If there are multiple patterns to detect, can a composite function detected
by one pattern be used as a part of a match for another
masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-578057856
ok I already have a working test case for detecting conv + bias + relu (Note
the "Composite" attribute). Now I really want to hook this
masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-577985906
ok reading your original proposal carefully I understand that I can look up
'Composite' attribute to know if a particular function has a
masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-577985906
ok reading your original proposal carefully I understand that I can look up
'Composite' attribute to know if a particular function has a
masahi edited a comment on issue #4771: [Relay] Added Merge Composite pass
URL: https://github.com/apache/incubator-tvm/pull/4771#issuecomment-577983004
From your test cases, I don't see why this pass would help remove pattern
matching on codegen side. Since Relay doesn't have