[Frontend] Refactor ServingTokenization entrypoint. (#46022)

Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io>
This commit is contained in:
wang.yuqi
2026-06-22 06:27:58 +00:00
committed by GitHub
parent 31124749d1
commit 1eb2cc961e
15 changed files with 322 additions and 302 deletions
@@ -12,7 +12,7 @@ from vllm.entrypoints.chat_utils import (
ChatTemplateContentFormatOption,
ConversationMessage,
)
from vllm.entrypoints.openai.engine.serving import RendererChatRequest, RendererRequest
from vllm.entrypoints.serve.engine.typing import RendererChatRequest, RendererRequest
from vllm.inputs import EngineInput, SingletonPrompt
from vllm.renderers import BaseRenderer, TokenizeParams, merge_kwargs
from vllm.renderers.inputs.preprocess import parse_model_prompt, prompt_to_seq
+2 -2
View File
@@ -52,7 +52,7 @@ class PoolingServingBase(ABC):
self.engine_client = engine_client
self.models = models
self.model_config = models.model_config
self.renderer = models.renderer
self.renderer = engine_client.renderer
self.vllm_config = engine_client.vllm_config
self.max_model_len = self.model_config.max_model_len
self.request_logger = request_logger
@@ -61,7 +61,7 @@ class PoolingServingBase(ABC):
self.chat_template_config = chat_template_config
# Shared thread pool executor for preprocessing and postprocessing.
self._executor: Executor = models.renderer._executor
self._executor: Executor = self.renderer._executor
self._preprocessing_async = make_async(
self._preprocessing, executor=self._executor
)