Skip to content

Navigation Menu

Appearance settings

View all features
- BY COMPANY SIZE
  Enterprises
  Small and medium teams
  Startups
  Nonprofits
- BY USE CASE
  App Modernization
  DevSecOps
  DevOps
  CI/CD
  View all use cases
- BY INDUSTRY
  Healthcare
  Financial services
  Manufacturing
  Government
  View all industries
View all solutions
- EXPLORE BY TOPIC
  AI
  Software Development
  DevOps
  Security
  View all topics
- EXPLORE BY TYPE
  Customer stories
  Events & webinars
  Ebooks & reports
  Business insights
  GitHub Skills
- SUPPORT & SERVICES
  Documentation
  Customer support
  Community forum
  Trust center
  Partners
View all resources
- COMMUNITY
  GitHub SponsorsFund open source developers
- PROGRAMS
  Security Lab
  Maintainer Community
  Accelerator
  GitHub Stars
  Archive Program
- REPOSITORIES
  Topics
  Trending
  Collections
- ENTERPRISE SOLUTIONS
  Enterprise platformAI-powered developer platform
- AVAILABLE ADD-ONS
  GitHub Advanced SecurityEnterprise-grade security features
  Copilot for BusinessEnterprise-grade AI features
  Premium SupportEnterprise-grade 24/7 support
Pricing

Search code, repositories, users, issues, pull requests...

Search

Clear

Search syntax tips

Provide feedback

We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Saved searches

Use saved searches to filter your results more quickly

Name

Query

To see all available qualifiers, see our documentation.

Appearance settings

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

NVIDIA / Model-Optimizer Public

Notifications You must be signed in to change notification settings
Fork 389
Star 2.6k

Code
Issues 56
Pull requests 143
Actions
Security and quality
Insights

Additional navigation options

Code
Issues
Pull requests
Actions
Security and quality
Insights

[OMNIML-3252][ONNX] Add real Q/DQ scales in Autotune#951

Merged

gcunhase merged 43 commits intoNVIDIA:mainNVIDIA/Model-Optimizer:mainfrom

gcunhase:dev/gcunhasergio/autotune_real_qdq_scalesgcunhase/TensorRT-Model-Optimizer:dev/gcunhasergio/autotune_real_qdq_scalesCopy head branch name to clipboard

Mar 11, 2026

Conversation Commits43 (43)Checks Files changed

Merged

[OMNIML-3252][ONNX] Add real Q/DQ scales in Autotune#951
gcunhase merged 43 commits intoNVIDIA:mainfrom
gcunhase:dev/gcunhasergio/autotune_real_qdq_scales

Commits

Commits on Mar 11, 2026

Initial autotune codebase
gcunhase
committed
Add more tests
gcunhase
committed
Refactor: PR #702
gcunhase
committed
Remove python path in tests
gcunhase
committed
Recover docstrings and simplify code (->, , )
gcunhase
committed
Added unittest for workflows.py (failing)
gcunhase
committed
Fix: 'Autotuning failed: 'PatternSchemes' object has no attribute 'node_inputs''
gcunhase
committed
Updated workflow test to test TRT and PythonTRT benchmarking
gcunhase
committed
Fix test: use_trtexec flag
gcunhase
committed
Add real scales to Q/DQ nodes
gcunhase
committed
fix precommit failures
gcunhase
committed
Fix: Add->Q/DQ->Activation(Relu)
gcunhase
committed
Fix: correctly dequantize Add input with shared Q/DQ
gcunhase
committed
[5916893] Fix weighted ops quantization logic: both input and weights Q/DQ need to be added or removed
gcunhase
committed
Changed keep_output_dir to True as default
gcunhase
committed
test_workflow was moved to 'tests/gpu/onnx'
gcunhase
committed
Removed cli.py, moved into __main__.py
gcunhase
committed
Removed PatternSchemes import from region_pattern.py: no longer needed.
gcunhase
committed
Added intermediate Autotune model to be removed at the end of the quantization workflow
gcunhase
committed
Removed _MUTATION_SPECS from autotuner.py: moved to autotuner_base.py
gcunhase
committed
Removed test_config and test_pattern_cache. Should be added in the original Auto Q/DQ PR.
gcunhase
committed
Fixed minor coderabbit suggestions
gcunhase
committed
Moved autotune imports to the top of the file
gcunhase
committed
Eliminate intermediate ONNX export in _find_nodes_to_quantize_autotune(). Directly use Insertion Points information.
gcunhase
committed
Add support for Add->Q/DQ->Relu patterns by including those 'Add' nodes in the nodes_to_quantize
gcunhase
committed
Add integration test
gcunhase
committed
Remove 'keep_output_dir' arg (no longer needed due to tmp path)
gcunhase
committed
Remove 'get_quantized_nodes' and other comments that are no longer needed
gcunhase
committed
Added docstring for 'default_dq_dtype' in workflows.py
gcunhase
committed
Added mode presets and additional autotune configurations
gcunhase
committed
Fixed tmp_path in test
gcunhase
committed
Fixed copilot comments
gcunhase
committed
Fix: skip rewiring in graph_utils if no index is found. This prevents silent corruption of the graph.
gcunhase
committed
Match args for preset mode default
gcunhase
committed
Exposed _StoreWithExplicitFlag
gcunhase
committed
Renamed new_ips to new_insertion_points
gcunhase
committed
Address coderabbit and copilot issues + other minor issues
gcunhase
committed
Address additional coderabbit and copilot issues
gcunhase
committed
Added real scales test in the integration workflow
gcunhase
committed
Address additional copilot issues: includes fix for op_types_to_quantize overwrite and other flags (should have the same behavior as pre-autotune)
gcunhase
committed
nit: added docstring and comment
gcunhase
committed
Created autotune utils
gcunhase
committed
Added 'bf16' as an option in workflows
gcunhase
committed

Footer

© 2026 GitHub, Inc.

Footer navigation

Terms
Privacy
Security
Status
Community
Docs
Contact

You can’t perform that action at this time.