test: Allow ensemble to create the final response even if some of the outputs are not created #7701
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
20 tasks
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
kthui
force-pushed
the
jacky-ensemble-output
branch
from
0706230 to
d40d8ff
Compare
kthui
changed the title
GuanLuo
approved these changes
Oct 15, 2024
rmccorm4
approved these changes
Oct 16, 2024
rmccorm4
reviewed
Oct 16, 2024
| @@ -177,6 +178,41 @@ def callback(start_time, result, error): | |||
| if expected_flags != response_flags: | |||
| self.assertTrue(False, "unexpeced sequence flags mismatch error") | |||
|
|
|||
| def test_ensemble_partial_add_sub(self): | |||
There was a problem hiding this comment.
I think we should have more testing in general to programatically track expected behavior of certain edge cases, such as:
- What if a composing model doesn't produce the outputs necessary to feed inputs of the next step?
- In an ensemble with outputs from 2 parallel steps, what if 1 step has outputs and the other step has no outputs?
- In an ensemble with outputs from 2 parallel steps, what if 1 step has outputs and the other step encounters an error? Is it considered OK to return the completed steps outputs, or is it an overall error?
- What if an ensemble doesn't produce any of its outputs?
- What if
These aren't immediately clear to me from scanning the code what would happen, so I think it would be good to enforce expected behavior via testing, and be able to catch if the behavior subtlely changes from future ensemble code changes
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Outdated suggestions cannot be applied.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.
What does the PR do?
Add test when the composing models of an ensemble do not provide all outputs on the model configuration. The inference should complete normally and return all outputs that are provided.
Checklist
<commit_type>: <Title>Commit Type:
Check the conventional commit type
box here and add the label to the github PR.
Related PRs:
triton-inference-server/core#399
Where should the reviewer start?
N/A
Test plan:
N/A
Caveats:
N/A
Background
N/A
Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)
N/A