Test ResNet Ops Support #28
Closed
Conversation
This file contains 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
### Changes Support e2m1 as type for weight compression. ### Reason for changes e2m1 compression. ### Related tickets CVS-140944
### Changes - Updated FastBiasCorrection algorithm to correct work with transposed MatMul layers. - Updated BiasCorrection algorithm to avoid recursion depth error. ### Reason for changes - FastBiasCorrection algorithm adaptation. - BiasCorrection improvement for many correctable layers. ### Related tickets - 144240 --------- Co-authored-by: Aleksei Kashapov <[email protected]>
### Changes Support TF 2.15 ### Reason for changes Support TF 2.15 ### Related tickets 133742 ### Tests pre-commit tests
### Changes - quantize addmm function - add gpt2 to post_training tests - add calculation int8 annd int4 const data for ptq ### Related tickets 142356
### Changes Remove reference data for TF 2.8 from the tests ### Reason for changes Follow-up openvinotoolkit#2609 ### Related tickets 133742 ### Tests pre-commit tests
…oolkit#2771) This reverts commit beb9c04. post_training_quantization/415 - running
### Changes The following functions were added: - numeric.py: eye - linalg.py: lstsq, svd ### Reason for changes Part of int4 with LoRA adaptation implementation. ### Related tickets CVS-135863 ### Tests tests/shared/test_templates/template_test_nncf_tensor.py
### Changes - https://numpy.org/devdocs/numpy_2_0_migration_guide.html - Enable numpy rules for ruff ### Related tickets 138867
### Changes - Updated FastBiasCorrection algorithm to correct work with transposed MatMul (OV) and Gemm (ONNX) layers. - Updated BiasCorrection algorithm to avoid recursion depth error. ### Reason for changes - FastBiasCorrection algorithm adaptation. - BiasCorrection improvement for many correctable layers. ### Related tickets - 144240 ### Tests - manual, post_training_quantization/418/ - passed
### Changes Changed default scale fot E2M1 weights from fp16 to fp8_e8m0 ### Reason for changes According to the "OCP Microscaling Formats (MX) Specification" ### Related tickets CVS-140944 ### Tests Will extend E2M1 tests.
github-actions
bot
added
documentation
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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
Added pytest to check if all the ops in the generated graph for resnet18 model are valid metatypes in NNCF
Reason for changes
Check if all model ops are supported by NNCF metatypes
Related tickets
openvinotoolkit#2778
Tests
New test file
test_models.pyadded intests/torch/fx/test_models.py