Rework HWConfig #3793
Merged
Rework HWConfig #3793
Conversation
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
github-actions
bot
added
documentation
AlexanderDokuchaev
requested review from
andrey-churkin,
daniil-lyakhov and
ljaljushkin
ljaljushkin
approved these changes
Jan 2, 2026
Contributor
ljaljushkin
left a comment
ljaljushkin
left a comment
There was a problem hiding this comment.
with the resolved comments from Daniil, LGTM
andrey-churkin
approved these changes
Jan 5, 2026
daniil-lyakhov
approved these changes
Jan 5, 2026
anzr299
added a commit
to anzr299/nncf
that referenced
this pull request
Jan 8, 2026
commit 7438f86 Author: Alexander Dokuchaev <[email protected]> Date: Wed Jan 7 15:43:50 2026 +0400 [PT] merge tests folders (openvinotoolkit#3821) Merge `tests/torch2` to `tests/torch` Removed unused code and data commit b7ba152 Author: Alexander Dokuchaev <[email protected]> Date: Wed Jan 7 14:17:02 2026 +0400 [CI] notebooks (openvinotoolkit#3813) - Format notebooks by pre-commit - Update profiller.py according [PyGuide.md](https://github.com/openvinotoolkit/nncf/blob/develop/docs/styleguide/PyGuide.md) - Set strict version in tools/requirements.txt commit 1bb59cf Author: Alexander Dokuchaev <[email protected]> Date: Wed Jan 7 14:13:55 2026 +0400 [doc] Ignored scope (openvinotoolkit#3818) Add documentation about using ignored scope. CVS-177621 commit 0c1054b Author: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Date: Wed Jan 7 14:10:19 2026 +0400 Bump astral-sh/setup-uv from 7.1.6 to 7.2.0 (openvinotoolkit#3831) Bumps [astral-sh/setup-uv](https://github.com/astral-sh/setup-uv) from 7.1.6 to 7.2.0. Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> commit 7ca2ad6 Author: Alexander Dokuchaev <[email protected]> Date: Tue Jan 6 14:16:02 2026 +0400 Copyrights 2026 (openvinotoolkit#3830) commit cdc19f8 Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 19:39:34 2026 +0400 Rework HWConfig (openvinotoolkit#3793) Replace JSON-based configuration loading with a Python-style HW Config. Simplifies modification and maintenance by keeping configuration logic in Python code instead of external JSON files. https://github.com/openvinotoolkit/nncf/actions/runs/20596514331 manual/job/post_training_quantization/767/ https://github.com/openvinotoolkit/nncf/actions/runs/20596794339 commit 2c4aaae Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 19:38:26 2026 +0400 [PT] prune + quantize (openvinotoolkit#3811) - Support quantization after pruning - Renamed `apply_magnitude_sparsity_binary_mask` to `apply_magnitude_binary_mask` - `wrap_model` don not raise error if model already wrapped - `find_const_node_in_constant_subgraph` can find weight after pruning operations CVS-171577 commit f1c33f8 Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 19:25:56 2026 +0400 nncf.build_graph (openvinotoolkit#3817) Replace `NNCFGraphFactory` to `nncf.build_graph` Suggestion of new api function `nncf.build_graph` to provide a simple way to get NNCFGraph for users. commit 1dc3d48 Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 19:16:38 2026 +0400 [PT] Build graph for unwrapped model (openvinotoolkit#3822) Add support of unwrapped model to build graph commit 89f4a53 Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 19:16:23 2026 +0400 Add test to validate version of package (openvinotoolkit#3825) To avoid merge auto-generated version. commit 9ee5c14 Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 15:56:25 2026 +0400 [CI] nightly tests on ubuntu (openvinotoolkit#3826) Run tests on ubuntu in nightly jobs Tests currently run on Ubuntu for pull requests and on a weekly schedule. This change ensures the develop branch is also continuously validated and remains stable. https://github.com/openvinotoolkit/nncf/actions/runs/20711605556 commit 329892c Author: Daniil Lyakhov <[email protected]> Date: Mon Jan 5 12:07:48 2026 +0100 [Collectors] Experimental collectors to common (openvinotoolkit#3761) * Redundant legacy tensor collectors are removed * Experimental tensor collectors moved to common * MyPy fixed for the collector classes * Some reducres/aggregators tests are extended * hash functions for some aggregators are fixed Reduce amount of redundant code, move common tensor collectors to common directory 150471 175176 openvinotoolkit#3041 commit bfe1095 Author: ruro <[email protected]> Date: Mon Jan 5 12:45:57 2026 +0300 Fix onnx Attention and torch SDPA quantization handling (openvinotoolkit#3751) - added `ONNXAttentionMetatype` for the opset 23 `Attention` ONNX node - fixed `scaled_dot_product_attention` quantization in `torch2` for the case when `Q`, `K` and `V` are parallel edges coming from the same input node See openvinotoolkit#3750 Fixes openvinotoolkit#3750 - `tests/onnx/quantization/test_graphs.py::test_synthetic_models_graph[AttentionModel]`  - `tests/torch2/function_hook/quantization/test_quantized_graphs.py::test_quantized_graphs[unbind_scaled_dot_product_attention_model]`  commit d21bbdd Author: Alexander Dokuchaev <[email protected]> Date: Mon Jan 5 11:52:51 2026 +0400 [PT] Add usage documentation about pruning (openvinotoolkit#3809) Add Usage.md Add api markers for relative function CVS-171577 --------- Co-authored-by: Lyalyushkin Nikolay <[email protected]> commit abf1dcd Author: Mostafa <[email protected]> Date: Mon Jan 5 09:34:44 2026 +0200 [NNCF] Add support for element-wise sign operation in tensor functions (openvinotoolkit#3824) Implemented `sign` function support in `src/nncf/tensor/functions/numeric.py`, `src/nncf/tensor/functions/numpy_numeric.py`, and `src/nncf/tensor/functions/torch_numeric.py`. The `sign` function is a fundamental element-wise operation required for various arithmetic manipulations within compression algorithms, including Half Quadratic Quantization. N/A The correctness of the changes was verified by adding a new test case `test_fn_sign` to `TemplateTestNNCFTensorOperators`. Fixes openvinotoolkit#3823 commit 0352237 Author: Daniil Lyakhov <[email protected]> Date: Mon Jan 5 08:01:30 2026 +0100 Happy New Copyright Year 2026🎉🎉🎉 (openvinotoolkit#3820) Copyrights are updated on 2026 year Year number incremented Alles gute in neuer Jahr!
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.
Changes
Replace JSON-based configuration loading with a Python-style HW Config.
Reason for changes
Simplifies modification and maintenance by keeping configuration logic in Python code instead of external JSON files.
Tests
https://github.com/openvinotoolkit/nncf/actions/runs/20596514331
manual/job/post_training_quantization/767/
https://github.com/openvinotoolkit/nncf/actions/runs/20596794339