[platforms] absorb worker cls difference into platforms folder (#10555)

Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
youkaichao
2024-11-21 21:00:32 -08:00
committed by GitHub
co-authored by Nick Hill
parent 446c7806b2
commit a111d0151f
21 changed files with 272 additions and 282 deletions
+1 -13
View File
@@ -1,4 +1,4 @@
from typing import Callable, List, Optional, Tuple, Type, Union
from typing import List, Optional, Union
from vllm.executor.executor_base import ExecutorAsyncBase
from vllm.executor.gpu_executor import GPUExecutor
@@ -6,7 +6,6 @@ from vllm.logger import init_logger
from vllm.model_executor.layers.sampler import SamplerOutput
from vllm.sequence import ExecuteModelRequest, PoolerOutput
from vllm.utils import make_async
from vllm.worker.worker_base import WorkerBase
logger = init_logger(__name__)
@@ -22,17 +21,6 @@ class XPUExecutor(GPUExecutor):
GPUExecutor._init_executor(self)
def _get_worker_module_and_class(
self) -> Tuple[str, str, Optional[Callable[[], Type[WorkerBase]]]]:
worker_class_fn = None
if self.speculative_config is not None:
raise NotImplementedError(
"XPU does not support speculative decoding")
else:
worker_module_name = "vllm.worker.xpu_worker"
worker_class_name = "XPUWorker"
return (worker_module_name, worker_class_name, worker_class_fn)
def execute_model(
self, execute_model_req: ExecuteModelRequest
) -> Optional[List[Union[SamplerOutput, PoolerOutput]]]: