xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-03-20 13:28:05 +08:00
Code Issues Packages Projects Releases Wiki Activity

Move docstring

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Pr0Wh1teGivee <calvin_zhu0210@outlook.com>
Signed-off-by: weichen <calvin_zhu0210@outlook.com>
This commit is contained in:
Harry Mellor 2025-12-05 11:03:38 +01:00 committed by weichen
parent 6413793466
commit ac674f6fc7
1 changed files with 4 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
vllm/config/scheduler.py
View File

@ -21,11 +21,7 @@ logger = init_logger(__name__)
RunnerType = Literal["generate", "pooling", "draft"]
SchedulerPolicy = Literal["fcfs", "priority", "sjf"]
""" SJF Scheduling Policy:
It stands for shortest-job-first — requests are scheduled by total prompt +
output length (shorter first), with aging to prevent starvation. For more
information, please check: https://github.com/vllm-project/vllm/issues/29406
"""
@config
@dataclass
@ -109,7 +105,9 @@ class SchedulerConfig:
- "fcfs" means first come first served, i.e. requests are handled in order
of arrival.\n
- "priority" means requests are handled based on given priority (lower
value means earlier handling) and time of arrival deciding any ties)."""
value means earlier handling) and time of arrival deciding any ties).\n
- "sjf" means shortest job first. Requests are scheduled by prompt length
(shortest first), with aging to prevent starvation."""
disable_chunked_mm_input: bool = False
"""If set to true and chunked prefill is enabled, we do not want to