xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-18 12:15:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

[multi-step] Raise error if not using async engine (#7703)

Browse Source
This commit is contained in:
William Lin 2024-08-21 11:49:19 -07:00 committed by GitHub
parent 1b32e02648
commit 91f4522cbf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/engine/llm_engine.py
View File

@ -1302,6 +1302,11 @@ class LLMEngine:
raise NotImplementedError( raise NotImplementedError(
"Pipeline parallelism is only supported through AsyncLLMEngine " "Pipeline parallelism is only supported through AsyncLLMEngine "
"as performance will be severely degraded otherwise.") "as performance will be severely degraded otherwise.")
if self.scheduler_config.num_scheduler_steps > 1:
raise NotImplementedError(
"Multiple scheduler steps (multi-step) are only supported "
"through AsyncLLMEngine. ")
seq_group_metadata_list, scheduler_outputs = self.scheduler[ seq_group_metadata_list, scheduler_outputs = self.scheduler[
0].schedule() 0].schedule()