-
Notifications
You must be signed in to change notification settings - Fork 528
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
support TF se_e2_a serialization; add a common test fixture to compare TF, PT, and DP models #3263
Conversation
…T, and DP models Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## devel #3263 +/- ##
==========================================
+ Coverage 75.27% 75.32% +0.04%
==========================================
Files 373 373
Lines 33252 33362 +110
Branches 1604 1604
==========================================
+ Hits 25031 25130 +99
- Misses 7350 7361 +11
Partials 871 871 ☔ View full report in Codecov by Sentry. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should the status of pt, tf, and dp in testing be considered equivalent? Should we set self.skip_tf, self.skip_pt, and self.skip_dp all? In future model implementations, there might be parts that are not available in tf as well.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I am not sure if the current design is a good idea, because to test the implementation of pt, one have to install the tf. suppose we have 5 backends, the env would be very heavy for the developers.
Ideally, the dp backend can be tested without any deep learning platforms, while any implementation, e.g. pt, can be tested without any other backend.
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
would it be possible to loop over a list of possible model parameters like
deepmd-kit/source/tests/pt/model/test_se_e2_a.py
Lines 43 to 45 in 392b9e0
for idt, prec in itertools.product( | |
[False, True], | |
["float64", "float32"], |
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
Signed-off-by: Jinzhe Zeng <[email protected]>
for more information, see https://pre-commit.ci
Signed-off-by: Jinzhe Zeng <[email protected]>
…into serialize-se-a
This reverts commit e192853.
Signed-off-by: Jinzhe Zeng <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like the design of decorator parameterized
very much.
Fix deepmodeling#3950. Backport a part of deepmodeling#3263 to r2 (the whole of deepmodeling#3263 is not likely to be backported). Signed-off-by: Jinzhe Zeng <[email protected]>
Fix #3950. Backport a part of #3263 to r2 (the whole of #3263 is not likely to be backported). <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Introduced new checks to ensure only weight and bias matrices are supported during tabulation operations. - **Refactor** - Removed redundant loop enforcing constraints on node keys related to weight and bias matrices to streamline operations. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Signed-off-by: Jinzhe Zeng <[email protected]>
Some codes are split from #2987.