Minor fix constant amax repr#1935
Open
mxinO wants to merge 2 commits into
Open
Conversation
Constant-amax quantizers intentionally omit the _amax buffer, but the representation treated every missing buffer as runtime-dynamic. Format the same E4M3 maximum used by the quantization path so diagnostics describe the active behavior. Constraint: Preserve true dynamic, static scalar, per-channel, and MX-format representations Rejected: Register a synthetic _amax buffer | would change quantizer state and calibration semantics for a display-only fix Confidence: high Scope-risk: narrow Reversibility: clean Directive: Keep constant-amax reporting derived from the value used by _get_amax Tested: tests/unit/torch/quantization/test_print.py (6 passed); focused Ruff, format, mypy, license, and Bandit hooks Not-tested: Full ModelOpt unit suite Signed-off-by: Meng Xin <mxin@nvidia.com>
A numeric value alone distinguishes constant mode from dynamic mode, but not from a stored calibrated scalar. Append a const marker so logs communicate both the effective value and its source without using per-channel bracket notation. Constraint: Preserve the existing per-channel [min, max](count) representation Rejected: Wrap the scalar in brackets | brackets already denote tensor ranges in quantizer diagnostics Confidence: high Scope-risk: narrow Reversibility: clean Tested: tests/unit/torch/quantization/test_print.py (6 passed); focused Ruff, format, mypy, license, and Bandit hooks Not-tested: Full ModelOpt unit suite Signed-off-by: Meng Xin <mxin@nvidia.com>
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
Contributor
📝 WalkthroughWalkthroughAdds a new conditional branch in ChangesConstant amax repr
Estimated code review effort: 1 (Trivial) | ~3 minutes Suggested reviewers: Quantization module maintainers familiar with 🚥 Pre-merge checks | ✅ 6✅ Passed checks (6 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
Contributor
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #1935 +/- ##
==========================================
- Coverage 77.69% 77.08% -0.61%
==========================================
Files 517 517
Lines 57535 57537 +2
==========================================
- Hits 44699 44350 -349
- Misses 12836 13187 +351
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (1)
modelopt/torch/quantization/nn/modules/tensor_quantizer.py (1)
1202-1213: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winUpdate docstring to document the constant-amax case.
The docstring's return list omits the new
(const)format now returned when_use_constant_amaxis set.📝 Suggested docstring update
def _short_amax(self, fmt=".2e"): """Short description of amax. Returns: 'dynamic': if _amax is not registered 'amax': if _amax is per-tensor '[min, max](size)': if _amax is per-channel + 'value(const)': if a constant amax is used (e.g. KV cache quantization) """🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@modelopt/torch/quantization/nn/modules/tensor_quantizer.py` around lines 1202 - 1213, The _short_amax method’s docstring is missing the constant-amax return case now handled by the _use_constant_amax branch. Update the return description in tensor_quantizer.TensorQuantizer._short_amax to document that it can return a constant-amax string with the “(const)” suffix, alongside the existing dynamic/per-tensor/per-channel cases, so the docstring matches the actual behavior.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Nitpick comments:
In `@modelopt/torch/quantization/nn/modules/tensor_quantizer.py`:
- Around line 1202-1213: The _short_amax method’s docstring is missing the
constant-amax return case now handled by the _use_constant_amax branch. Update
the return description in tensor_quantizer.TensorQuantizer._short_amax to
document that it can return a constant-amax string with the “(const)” suffix,
alongside the existing dynamic/per-tensor/per-channel cases, so the docstring
matches the actual behavior.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 8fbe8244-001c-42da-bf64-e143ab8c72e6
📒 Files selected for processing (2)
modelopt/torch/quantization/nn/modules/tensor_quantizer.pytests/unit/torch/quantization/test_print.py
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
1 participant
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.
What does this PR do?
Type of change: ?
The constant amax in kv cache is showing
dynamicwhich is confusing, making it show the "value(const)".Usage
# Add a code snippet demonstrating how to use thisTesting
Before your PR is "Ready for review"
Make sure you read and follow Contributor guidelines and your commits are signed (
git commit -s -S).Make sure you read and follow the Security Best Practices (e.g. avoiding hardcoded
trust_remote_code=True,torch.load(..., weights_only=False),pickle, etc.).CONTRIBUTING.md: ✅ / ❌ / N/AAdditional Information
Summary by CodeRabbit
Bug Fixes
Tests