[Unity] Add R.quantize and R.dequantize ops#15849
Merged
masahi merged 1 commit intoapache:unityfrom Oct 3, 2023
Merged
Conversation
Contributor
Author
ibsidorenko
changed the title
masahi
reviewed
Oct 3, 2023
Member
masahi
left a comment
masahi
left a comment
There was a problem hiding this comment.
Also needs to fix the conflict.
python/tvm/relax/op/qdq.py
Outdated
| The output scale. | ||
|
|
||
| zero_point : tvm.relay.Expr | ||
| The output zero_point. |
Member
There was a problem hiding this comment.
I think "input" scale / zp would be more correct?
src/relax/op/tensor/qdq.cc
Outdated
| } | ||
|
|
||
| // Check datatype of zero_point param: | ||
| if (zp_sinfo->dtype != DataType::Int(32)) { |
Member
There was a problem hiding this comment.
I remember that zp needs to be 8 bit in practice. How about enforcing int8 here?
Contributor
Author
There was a problem hiding this comment.
I think it is possible. I have never met the cases when zero_point is out of [-128, 127] range.
|
|
||
| /*! | ||
| * \file src/relax/op/tensor/qdq.h | ||
| * \brief The functions to make Relax quantize/dequantize operator calls. |
Contributor
Author
There was a problem hiding this comment.
There's no need for that in this PR. But it was added for the future when we want to construct quantize/dequantize op in one of TVM passes.
This commit adds 2 new operations (R.quantize and R.dequantize) and supports them in LegalizeOps pass.
masahi
approved these changes
Oct 3, 2023
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This commit adds 2 new operations (
R.quantizeandR.dequantize) and supports them inLegalizeOpspass.