-
Notifications
You must be signed in to change notification settings - Fork 2
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
[V1] Various updates #28
[V1] Various updates #28
Conversation
omit_defaults=True, | ||
gc=False): |
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.
These also have performance benefit
vllm/v1/engine/async_llm.py
Outdated
def _finish_stream(self, request_id: str): | ||
stream = self.request_streams.pop(request_id) | ||
if stream is not None: | ||
stream.finish() |
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.
We need to remove the stream if it finishes due to a finished request output or due to client cancellation.
vllm/v1/engine/async_llm.py
Outdated
await self.engine_core.abort_requests_async(request_ids) | ||
self.detokenizer.abort_requests(request_ids) |
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.
This shouldn't be needed because the detokenizer removes finished requests itself.
encoder.encode_into(outputs, buffer) | ||
socket.send_multipart((buffer, ), |
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.
This will reuse the buffer rather than allocating one each time.
See inline comments