Fix quantization for input to reference model #2317
Closed
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/2317
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (3 Unrelated Failures)As of commit f1b2bf8 with merge base f9cad4e ( FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
per
added
the
partner: arm
|
Will fix the double decorators causing the failures. |
per
commented
Mar 8, 2024
backends/arm/test/ops/test_add.py
Outdated
| self._test_add_tosa_BI_pipeline(self.Add2(), test_data) | ||
|
|
||
| @unittest.skipIf( | ||
| not VELA_INSTALLED, | ||
| "There is no point in running U55 tests if the Vela tool is not installed", | ||
| ) | ||
| def test_add2_u55_BI(self): | ||
| test_data = (torch.ones(1, 1, 4, 4), torch.ones(1, 1, 4, 1)) | ||
| @parameterized.expand(Add2.test_parameters) |
There was a problem hiding this comment.
@parameterized and @UnitTest causing failures.
per
force-pushed
the
reference_model_output
branch
from
eac5f44 to
63bd288
Compare
per
force-pushed
the
reference_model_output
branch
3 times, most recently
from
c9ff559 to
8b51570
Compare
Add the zerpoint instead of subtracting. This worked since the tests so far used the ones as inputs which quantize to a zp of -128 which gives the same np.int8 result in both cases since the int8 wraps. Also needs to round and clip the scaled values to the int8 range. Signed-off-by: Per Åstrand <[email protected]> Change-Id: Ideaed6d072a4065573b38fb7476c7dbe8ba814fd
Fix order of decorators to expand unittest first, and then parameterized input. Fix bug in add operator conversion to handle different scales correctly. Signed-off-by: Per Åstrand <[email protected]> Change-Id: Ic228cf0215e8171392776739936a53c025802fd5
per
force-pushed
the
reference_model_output
branch
from
8b51570 to
f1b2bf8
Compare
digantdesai
approved these changes
Mar 21, 2024
backends/arm/test/ops/test_add.py
Outdated
| self._test_add_tosa_BI_pipeline(self.Add2(), test_data) | ||
|
|
||
| @unittest.skipIf( | ||
| not VELA_INSTALLED, | ||
| "There is no point in running U55 tests if the Vela tool is not installed", | ||
| ) | ||
| def test_add2_u55_BI(self): | ||
| test_data = (torch.ones(1, 1, 4, 4), torch.ones(1, 1, 4, 1)) | ||
| @parameterized.expand(Add2.test_parameters) |
|
@digantdesai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
@digantdesai merged this pull request in d06ccd2. |
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.
Add the zerpoint instead of subtracting. This worked since the tests so far used the ones as inputs which quantize to a zp of -128 which gives the same np.int8 result in both cases since the int8 wraps. Also needs to round and clip the scaled values to the int8 range.
Signed-off-by: Per Åstrand [email protected]