[Core] Eliminate parallel worker per-step task scheduling overhead (#4894)
This commit is contained in:
@@ -1,7 +1,7 @@
|
||||
import importlib
|
||||
import os
|
||||
from abc import ABC, abstractmethod
|
||||
from typing import Dict, List, Set, Tuple
|
||||
from typing import Dict, List, Optional, Set, Tuple
|
||||
|
||||
from vllm.logger import init_logger
|
||||
from vllm.lora.request import LoRARequest
|
||||
@@ -48,8 +48,9 @@ class WorkerBase(ABC):
|
||||
|
||||
@abstractmethod
|
||||
def execute_model(
|
||||
self,
|
||||
execute_model_req: ExecuteModelRequest) -> List[SamplerOutput]:
|
||||
self,
|
||||
execute_model_req: Optional[ExecuteModelRequest] = None
|
||||
) -> List[SamplerOutput]:
|
||||
"""Executes at least one model step on the given sequences, unless no
|
||||
sequences are provided."""
|
||||
raise NotImplementedError
|
||||
|
||||
Reference in New Issue
Block a user