xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 05:15:42 +08:00
Code Issues Packages Projects Releases Wiki Activity

[V1][Minor] Minor enhancements on scheduler (#14732)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon 2025-03-13 08:53:22 -07:00 committed by GitHub
parent f53a0586b9
commit 01b3fd0af7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 3 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
vllm/v1/core/scheduler.py
View File

@ -587,9 +587,6 @@ class Scheduler:
if spec_token_ids is not None:
request.spec_token_ids = spec_token_ids[req_index]
# Get prompt logprobs for this request.
prompt_logprobs_tensors = prompt_logprobs_dict.get(req_id)
stopped = False
new_logprobs = None
new_token_ids: list[int] = []
@ -622,6 +619,8 @@ class Scheduler:
new_token_ids,
)
# Get prompt logprobs for this request.
prompt_logprobs_tensors = prompt_logprobs_dict.get(req_id)
# Transmit partial if chunked prefill & prompt logprobs is enabled
if new_token_ids or prompt_logprobs_tensors is not None:
# Add EngineCoreOutput for this Request.
@ -693,8 +692,7 @@ class Scheduler:
if request.status == RequestStatus.RUNNING:
self.running.remove(request)
if request.request_id in self.scheduled_req_ids:
self.scheduled_req_ids.remove(request.request_id)
self.scheduled_req_ids.discard(request.request_id)
else:
self.waiting.remove(request)
request.status = finished_status