Expose priority parameter in Engine.generate() and Engine.async_generate() (#18944)

Signed-off-by: PeaBrane <peabrane@gmail.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Yan Ru Pei
2026-02-17 19:14:22 -08:00
committed by GitHub
parent 34d975b18f
commit 83e24e2eb4
3 changed files with 7 additions and 0 deletions

View File

@@ -30,6 +30,7 @@ class EngineBase(ABC):
bootstrap_room: Optional[Union[List[int], int]] = None,
data_parallel_rank: Optional[int] = None,
rid: Optional[Union[List[str], str]] = None,
priority: Optional[int] = None,
) -> Union[Dict, Iterator[Dict]]:
"""Generate outputs based on given inputs."""
pass

View File

@@ -236,6 +236,7 @@ class Engine(EngineBase):
external_trace_header: Optional[Dict] = None,
rid: Optional[Union[List[str], str]] = None,
session_params: Optional[Dict] = None,
priority: Optional[int] = None,
) -> Union[Dict, Iterator[Dict]]:
"""
The arguments of this function is the same as `sglang/srt/managers/io_struct.py::GenerateReqInput`.
@@ -274,6 +275,7 @@ class Engine(EngineBase):
external_trace_header=external_trace_header,
rid=rid,
session_params=session_params,
priority=priority,
)
generator = self.tokenizer_manager.generate_request(obj, None)
@@ -325,6 +327,7 @@ class Engine(EngineBase):
external_trace_header: Optional[Dict] = None,
rid: Optional[Union[List[str], str]] = None,
session_params: Optional[Dict] = None,
priority: Optional[int] = None,
) -> Union[Dict, AsyncIterator[Dict]]:
"""
The arguments of this function is the same as `sglang/srt/managers/io_struct.py::GenerateReqInput`.
@@ -364,6 +367,7 @@ class Engine(EngineBase):
external_trace_header=external_trace_header,
rid=rid,
session_params=session_params,
priority=priority,
)
generator = self.tokenizer_manager.generate_request(obj, None)

View File

@@ -108,6 +108,7 @@ class HttpServerEngineAdapter(EngineBase):
token_ids_logprob=None,
lora_path=None,
custom_logit_processor=None,
priority=None,
):
payload = {
"text": prompt,
@@ -120,6 +121,7 @@ class HttpServerEngineAdapter(EngineBase):
"token_ids_logprob": token_ids_logprob,
"lora_path": lora_path,
"custom_logit_processor": custom_logit_processor,
"priority": priority,
}
# Filter out None values
payload = {k: v for k, v in payload.items() if v is not None}