TrussServer supports request/repsonse #1148
Merged
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
marius-baseten
force-pushed
the
marius/requests
branch
3 times, most recently
from
2c10e41 to
eafad4e
Compare
marius-baseten
force-pushed
the
marius/requests
branch
from
eafad4e to
335bbbe
Compare
squidarth
reviewed
Sep 18, 2024
marius-baseten
force-pushed
the
marius/requests
branch
from
25e67cf to
9d3e562
Compare
squidarth
reviewed
Sep 18, 2024
There was a problem hiding this comment.
this is AWESOME. Very excited for these changes -- particularly enabling SSE (which we'll get on day 1), and the stacktrace change (this has been bothering me for a long time, glad we finally got around to it).
took a first pass w a couple questions, will take a closer look later
| @@ -974,3 +874,275 @@ def enable_gpu_fn(conf): | |||
| # But make sure traces have parents at all. | |||
| assert len(user_parents) > 3 | |||
| assert len(truss_parents) > 3 | |||
|
|
|||
|
|
|||
| # Returning Response Objects ########################################################### | |||
There was a problem hiding this comment.
One thought -- do you think it makes sense to have an integration test for the stack trace we made in this PR?
There was a problem hiding this comment.
I was too lazy to lookup the traceback API and the initial implementation was totally bloated by chat GPT - I had this feeling that it could be simplified, now it's pretty straightforward.
But adding a test would be good too!
marius-baseten
force-pushed
the
marius/requests
branch
5 times, most recently
from
aa764b8 to
ef153e4
Compare
marius-baseten
force-pushed
the
marius/requests
branch
from
ef153e4 to
c5e0d36
Compare
squidarth
approved these changes
Sep 19, 2024
marius-baseten
force-pushed
the
marius/requests
branch
from
7d39fde to
ba1de60
Compare
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
Support access to requests object in truss model and allow returning response objects. This follows the internal proposal.
Includes:
ModelDefinitionErroris raised.truss_server.pyinference_server.pyandmodel_wrapper.py:truss_servercontains the actual server implementation, as an application entrypoint a super shallowmain.pyfile is used.ModelDescriptorthat is just initialized once and contains information whether methods are present and sync/async etc.Side quests:
intercept_exceptionsfunction wrapper with contextmanager so it can be easier used on non-function code sections.importlib.util.module_from_specinstead ofimportlib.import_modulebecause it has side effects in unit/integration tests.model_wrapperloading. - Root cause for prior failure was import side effects (previous point).test_model_wrapper_streaming_timeoutthat had missing@pytest.mark.anyio.This PR includes changes from #1103, which would be obsolete.
💻 How
🔬 Testing