[op conformance] Fix threshold bugs in ov::test::utils::compare #22972
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
github-actions
bot
added
the
category: IE Tests
sbalandi
force-pushed
the
accuracy1
branch
3 times, most recently
from
1ef76c6 to
828e681
Compare
sbalandi
changed the title
iefode
reviewed
Feb 22, 2024
iefode
approved these changes
Feb 23, 2024
|
@eshoguli FYI |
Merged
sbalandi
added a commit
to sbalandi/openvino
that referenced
this pull request
Feb 28, 2024
…re (openvinotoolkit#22972)" This reverts commit 1e5d0e5.
sbalandi
added a commit
to sbalandi/openvino
that referenced
this pull request
Feb 28, 2024
…re (openvinotoolkit#22972)" This reverts commit 1e5d0e5.
github-merge-queue bot
pushed a commit
that referenced
this pull request
May 16, 2024
…NN deconv OP from conv_backward_data (#23596) ### Details: - *Re-enable conv with brgemm implement on avx2 platform with fp32 , also heuristic cause by wavernn-upsampler regression* - *Switch OV cpu deconv implement from ONEDNN convbackward_data primitive to deconv primitive. Support const and non-const weight; Always expose planar weight layout to cpu graph; weight cache to save blocked const weight ; runtime reorder for none-const weight* - *enable bias fusing for fp32 to replace legacy depthwise fusing* - *enable deconv brgconv for fp32 and INT8 deconv* - *update test case* ### About test case/WA update: - *OV has feature to set the output shape of conv_backward_data. Which would cause pad_r < 0 && pad_r + dil < 0, so applied the WA in onednn to create deconv OP* - *Deconv only has one ref entry(CPU_INSTANCE(ref_deconvolution_fwd_t) for JIT,gemm and c++ reference implement. So one layout can't use RTinfo to filter implement. Update test cases with 1x1 related or SSE related. For example: backward data has 2 entries when input is nChw16C : jit_avx512_common_1x1_convolution_bwd_data_f32_t jit_avx512_common_convolution_bwd_data_t<f32> to support So OV filter can work to get desired implement type. However, deconv only has only, it means that nChw16C would only get one implement type. So filter would not work.* - *Change stride to 1 unit because of brgemm limitation* - *subgraph concat deconv test can't hit ref implement because the gemm and ref share one entry implement.* - *Conformance test conv related failed, but can pass when merging 3.3 with brg. Failed caused by . Seems threshold changes. Also AMX and avv512 core would fail on these conformance test but not enabled on Azure platforms. Guess the expected result is based on avx2 jit. So current just remove these test* ### Tickets: - *CVS-133120, CVS-122688*
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.
Details:
Tickets: