xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-11 10:55:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Convert use_structured_output property into constant (#23060)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon 2025-08-17 12:41:38 -07:00 committed by GitHub
parent c55bc1db26
commit 6d243efeda
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
vllm/v1/request.py
View File

@ -54,8 +54,7 @@ class Request:
time.time() time.time()
self.status = RequestStatus.WAITING self.status = RequestStatus.WAITING
if sampling_params and sampling_params.guided_decoding is not None: self.use_structured_output = False
self.status = RequestStatus.WAITING_FOR_FSM
self.events: list[EngineCoreEvent] = [] self.events: list[EngineCoreEvent] = []
self.stop_reason: Union[int, str, None] = None self.stop_reason: Union[int, str, None] = None
@ -63,12 +62,15 @@ class Request:
self.kv_transfer_params: Optional[dict[str, Any]] = None self.kv_transfer_params: Optional[dict[str, Any]] = None
if pooling_params is not None: if pooling_params is not None:
# Pooling models.
self.max_tokens = 1 self.max_tokens = 1
elif sampling_params is not None: elif sampling_params is not None:
# Generative models.
assert sampling_params.max_tokens is not None assert sampling_params.max_tokens is not None
self.max_tokens = sampling_params.max_tokens self.max_tokens = sampling_params.max_tokens
if sampling_params.guided_decoding is not None: if sampling_params.guided_decoding is not None:
self.status = RequestStatus.WAITING_FOR_FSM self.status = RequestStatus.WAITING_FOR_FSM
self.use_structured_output = True
if sampling_params.extra_args is not None: if sampling_params.extra_args is not None:
self.kv_transfer_params = \ self.kv_transfer_params = \
@ -192,11 +194,6 @@ class Request:
num_tokens = self.mm_positions[input_id].length num_tokens = self.mm_positions[input_id].length
return num_tokens return num_tokens
@property
def use_structured_output(self) -> bool:
return self.sampling_params is not None and \
self.sampling_params.guided_decoding is not None
def record_event( def record_event(
self, self,
event_type: EngineCoreEventType, event_type: EngineCoreEventType,