xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-18 06:15:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] get_and_reset only when scheduler outputs are not empty (#6266)

Browse Source
This commit is contained in:
Mor Zusman 2024-07-11 17:40:20 +03:00 committed by GitHub
parent 8a1415cf77
commit 55f692b46e
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/engine/async_llm_engine.py
View File

@ -225,11 +225,11 @@ class _AsyncLLMEngine(LLMEngine):
""" """
seq_group_metadata_list, scheduler_outputs = self.scheduler[ seq_group_metadata_list, scheduler_outputs = self.scheduler[
virtual_engine].schedule() virtual_engine].schedule()
finished_requests_ids = self.scheduler[
virtual_engine].get_and_reset_finished_requests_ids()
if not scheduler_outputs.is_empty(): if not scheduler_outputs.is_empty():
# Execute the model. # Execute the model.
finished_requests_ids = self.scheduler[
virtual_engine].get_and_reset_finished_requests_ids()
execute_model_req = ExecuteModelRequest( execute_model_req = ExecuteModelRequest(
seq_group_metadata_list=seq_group_metadata_list, seq_group_metadata_list=seq_group_metadata_list,
blocks_to_swap_in=scheduler_outputs.blocks_to_swap_in, blocks_to_swap_in=scheduler_outputs.blocks_to_swap_in,

4
vllm/engine/llm_engine.py
View File

@ -871,10 +871,10 @@ class LLMEngine:
"as performance will be severely degraded otherwise.") "as performance will be severely degraded otherwise.")
seq_group_metadata_list, scheduler_outputs = self.scheduler[ seq_group_metadata_list, scheduler_outputs = self.scheduler[
0].schedule() 0].schedule()
finished_requests_ids = self.scheduler[
0].get_and_reset_finished_requests_ids()
if not scheduler_outputs.is_empty(): if not scheduler_outputs.is_empty():
finished_requests_ids = self.scheduler[
0].get_and_reset_finished_requests_ids()
execute_model_req = ExecuteModelRequest( execute_model_req = ExecuteModelRequest(
seq_group_metadata_list=seq_group_metadata_list, seq_group_metadata_list=seq_group_metadata_list,
blocks_to_swap_in=scheduler_outputs.blocks_to_swap_in, blocks_to_swap_in=scheduler_outputs.blocks_to_swap_in,