diff --git a/vllm/config/scheduler.py b/vllm/config/scheduler.py index bb0ecf38a7468..a9640943b8c18 100644 --- a/vllm/config/scheduler.py +++ b/vllm/config/scheduler.py @@ -106,7 +106,7 @@ class SchedulerConfig: of arrival.\n - "priority" means requests are handled based on given priority (lower value means earlier handling) and time of arrival deciding any ties).\n - - "sjf" means shortest job first. Requests are scheduled by prompt length + - "sjf" means shortest job first. Requests are scheduled by prompt length (shortest first), with aging to prevent starvation.""" disable_chunked_mm_input: bool = False