refactor(runner): add InferenceError to all pipelines #188
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
This commit adds the inference error logic from the SAM2 pipeline to all pipelines so users are given a warning when they supply wrong arguments.
rickstaa
force-pushed
the
add_all_pipelines_inference_error
branch
from
f08aced
to
a469892
Compare
This commit ensures that users get a descriptive error message when the GPU runs out of memory.
This commit ensures that all response errors are known by FastAPI and therefore shown in the docs.
This commit adds some missing error handling to the pipeline worker functions.
rickstaa
commented
Sep 22, 2024
This commit improves the out of memory error handling by using the native torch error.
rickstaa
force-pushed
the
add_all_pipelines_inference_error
branch
from
dba6dc8
to
53d76eb
Compare
This commit ensures that errors thrown by the runner are forwarded to the orchestrator. It applies the logic used by the SAM2 and audio-to-text pipelines to the other pipelines.
This commit applies the black formatter to the PR files.
This commit removes the redundant str call.
leszko
approved these changes
Oct 14, 2024
runner/app/routes/image_to_video.py
Outdated
| Returns: | ||
| A JSONResponse with the appropriate error message and status code. | ||
| """ | ||
| if isinstance(e, torch.cuda.OutOfMemoryError): |
There was a problem hiding this comment.
I see this part of the code is repeated in every pipeline, would it maybe make sense to extract it and reuse?
There was a problem hiding this comment.
@leszko thanks for the quick review. Yea I was planning to extract it similar to the logic @mjh1 implemented in #226 (See #226 (comment)). I however wanted to do this in a subsequent pull request. I can do it here and co-author @mjh1 👍🏻.
There was a problem hiding this comment.
@leszko as also stated in the discord I added a variant of @gioelecerati's logic in ae8df74.
This commit introduces a global error handling configuration and function to streamline error management across different pipelines. The new `handle_pipeline_exception` function centralizes error handling logic, allowing pipelines to override it if necessary. This change reduces code duplication and improves maintainability. Co-authored-by: rickstaa <rick.staa@outlook.com>
rickstaa
force-pushed
the
add_all_pipelines_inference_error
branch
from
21a3d77
to
ae8df74
Compare
This commit ensures that pipelines can overwrite the default error message when the Global error configuration contains a empty string.
This commit adds a test for the 'handle_pipeline_exception' route utility function. It also fixes some errors into that function.
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.
This pull request adds the inference error logic from the SAM2 pipeline to all pipelines so users are given a warning when they supply wrong arguments. Please note that the 422 thrown by FastAPI are not handled corretly and we don't do validation at the go-livepeer side yet (see https://linear.app/livepeer-ai/project/improve-error-handling-117c11ff78f3/issues). As a result although this pull request improves the behavoir we don't yet have an optimised solution.