Preserve shape ops when removing QDQ pairs#5048
Open
ikalinic wants to merge 6 commits into
Open
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## develop #5048 +/- ##
===========================================
+ Coverage 92.78% 92.81% +0.03%
===========================================
Files 596 597 +1
Lines 32011 32164 +153
===========================================
+ Hits 29700 29851 +151
- Misses 2311 2313 +2
🚀 New features to boost your workflow:
|
pfultz2
requested changes
Jul 8, 2026
Regressions detected 🔴 |
|
TedThemistokleous
self-requested a review
July 16, 2026 14:15
TedThemistokleous
requested changes
Jul 16, 2026
TedThemistokleous
left a comment
TedThemistokleous
left a comment
Collaborator
There was a problem hiding this comment.
Hey thanks for your contribution!
I have a few suggested changes and some initial questions
| inline auto get_input_path(instruction_ref ins) | ||
| { | ||
| return unfold(ins, [](instruction_ref x) -> std::optional<instruction_ref> { | ||
| if(x->inputs().size() != 1) |
Collaborator
There was a problem hiding this comment.
What if there's multiple inputs?
Is this intending on check for single inputs or non zero non multi inputs. Some ops can have two inputs like gather (calculate shape for indicies and data).
| if(q_pos == input_path.end()) | ||
| return std::nullopt; | ||
| auto is_pack_unpack = [](auto x) { | ||
| return contains({"pack_fp4", "unpack_fp4"}, x->name()); |
Collaborator
There was a problem hiding this comment.
Make this a helper instead of a lamda so we can target all pack/unpacks in general.
Do the same for has_pack_unpack below
| if(contains(replay_ops(), x->name())) | ||
| { | ||
| if(not has_pack_unpack) | ||
| ops.push_back(x->get_operator()); |
Collaborator
There was a problem hiding this comment.
I think we should separate this between "should we add" and "we are adding the replay op" to the list. Is the intent here to check and ensure we have replays that we can then if not just return no replacement?
| if(replacement->get_shape() != dq_ins->get_shape()) | ||
| return std::nullopt; | ||
| return replacement; | ||
| } |
Collaborator
There was a problem hiding this comment.
There's a few code paths here where things just go to nullopt after a bunch of operations. Is there a way to reduce the complexity of this or return earlier by bundling a few of these checks?
| static const std::unordered_set<std::string> ops = { | ||
| "pack_fp4", "unpack_fp4", "broadcast", "slice", "reshape", "reshape_lazy", "pad"}; | ||
| return ops; | ||
| } |
Collaborator
There was a problem hiding this comment.
Not sure why this had to change to a function same with below for replay_ops return this back as a regular static const std:unordered_set for both.
| } | ||
| if(not has_pack_unpack) | ||
| { | ||
| std::reverse(ops.begin(), ops.end()); |
Collaborator
There was a problem hiding this comment.
Not sure why we need to do a reverse here?, also we're inserting new items in the graph with the insert below.
Collaborator
|
Adding Shiv since he's looked a lot of the QDQ stuff |
| new_cat_inputs.end(), | ||
| replacement_inputs.begin(), | ||
| [](const auto& input) { return *input; }); | ||
| m.replace_instruction(ins, cat_ins->get_operator(), replacement_inputs); |
Contributor
There was a problem hiding this comment.
Can you add a regression test for this finder? So its easier to understand what was failing before and is now fixed by this change
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
4 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.
Motivation
Fix
simplify_qdqso Q/DQ pair removal does not accidentally drop intervening shape operations or produce shape-changing rewrites.Technical Details
Updated Q/DQ removal to replay supported shape operations when removing non-FP4 fake quantization chains, while continuing to drop FP4 pack/unpack adapter operations where appropriate. Added guards for concat QLinear splitting when sliced scale or zero-point shapes do not match the corresponding concat input. Added a regression test for preserving
sliceandreshapethrough Q/DQ removal.Changelog Category
Add a
CHANGELOG.mdentry for any option other thanNot Applicable