nvidia-trt: no stop request, stop response for decoupled model #17764
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
- detect decoupled model and request explicit stop response
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎ 1 Ignored Deployment
|
Closed
mkhludnev
commented
Feb 19, 2024
| @@ -218,7 +218,7 @@ def _invoke_triton(self, model_name, inputs, outputs, stop_words): | |||
| result_queue = StreamingResponseGenerator( | |||
| self, | |||
| request_id, | |||
| force_batch=False, | |||
| signal_stop=False, | |||
There was a problem hiding this comment.
renaming to meaningful. Inverting it at line 405.
TLDR triton doesn't let one cancel a request triton-inference-server/server#4818
| @@ -238,6 +238,7 @@ def _invoke_triton(self, model_name, inputs, outputs, stop_words): | |||
| inputs=inputs, | |||
| outputs=outputs, | |||
| request_id=request_id, | |||
| enable_empty_final_response=self.client.get_model_config(model_name).config.model_transaction_policy.decoupled | |||
There was a problem hiding this comment.
I want to improve it here. Maybe cache it or use it as model_ready call above. I also don't know how much this grpc code is null proof.
dosubot
bot
added
size:M
|
Should be against the https://github.com/langchain-ai/langchain-nvidia repo instead! |
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Outdated suggestions cannot be applied.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.
Description:
Issue: now it sends
stop:Trueas an input, and triton rejects it; then withouttriton_enable_empty_final_response=Truestream hangs foreverAdd tests and docs: If you're adding a new integration, please include
docs/docs/integrationsdirectory.Lint and test: Run
make format,make lintandmake testfrom the root of the package(s) you've modified. See contribution guidelines for more: https://python.langchain.com/docs/contributing/If no one reviews your PR within a few days, please @-mention one of baskaryan, efriis, eyurtsev, hwchase17.