biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[2/N] executor pass the complete config to worker/modelrunner (#9938)

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
youkaichao
2024-11-02 07:35:05 -07:00
committed by GitHub
parent 1d4cfe2be1
commit e893795443
44 changed files with 249 additions and 579 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
vllm/engine/async_llm_engine.py
View File

@@ -7,8 +7,8 @@ from typing import (Any, AsyncGenerator, Callable, Coroutine, Dict, Iterable,
from weakref import ReferenceType
import vllm.envs as envs
from vllm.config import (DecodingConfig, EngineConfig, LoRAConfig, ModelConfig,
ParallelConfig, SchedulerConfig)
from vllm.config import (DecodingConfig, LoRAConfig, ModelConfig,
ParallelConfig, SchedulerConfig, VllmConfig)
from vllm.core.scheduler import SchedulerOutputs
from vllm.engine.arg_utils import AsyncEngineArgs
from vllm.engine.async_timeout import asyncio_timeout
@@ -604,7 +604,7 @@ class AsyncLLMEngine(EngineClient):
@classmethod
def _get_executor_cls(
cls, engine_config: EngineConfig) -> Type[ExecutorAsyncBase]:
cls, engine_config: VllmConfig) -> Type[ExecutorAsyncBase]:
distributed_executor_backend = (
engine_config.parallel_config.distributed_executor_backend)
if isinstance(distributed_executor_backend, type):
@@ -663,7 +663,7 @@ class AsyncLLMEngine(EngineClient):
def from_engine_args(
cls,
engine_args: AsyncEngineArgs,
engine_config: Optional[EngineConfig] = None,
engine_config: Optional[VllmConfig] = None,
start_engine_loop: bool = True,
usage_context: UsageContext = UsageContext.ENGINE_CONTEXT,
stat_loggers: Optional[Dict[str, StatLoggerBase]] = None,