[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
+11 -6
View File
@@ -34,7 +34,7 @@ from vllm.entrypoints.openai.models.serving import OpenAIServingModels
from vllm.entrypoints.serve.elastic_ep.middleware import ScalingMiddleware
from vllm.entrypoints.serve.render.serving import OpenAIServingRender
from vllm.entrypoints.serve.sagemaker.api_router import sagemaker_standards_bootstrap
from vllm.entrypoints.serve.tokenize.serving import OpenAIServingTokenization
from vllm.entrypoints.serve.tokenize.serving import ServingTokenization
from vllm.entrypoints.serve.utils.api_utils import (
cli_env_setup,
log_non_default_args,
@@ -376,8 +376,7 @@ async def init_app_state(
log_error_stack=args.log_error_stack,
)
state.openai_serving_tokenization = OpenAIServingTokenization(
engine_client,
state.serving_tokenization = ServingTokenization(
state.openai_serving_models,
state.openai_serving_render,
request_logger=request_logger,
@@ -461,9 +460,15 @@ async def init_render_app_state(
)
state.openai_serving_models = model_registry
# Expose tokenization via the render handler (no engine required).
state.openai_serving_tokenization = state.openai_serving_render
state.serving_tokenization = ServingTokenization(
model_registry,
state.openai_serving_render,
request_logger=request_logger,
chat_template=resolved_chat_template,
chat_template_content_format=args.chat_template_content_format,
default_chat_template_kwargs=args.default_chat_template_kwargs,
trust_request_chat_template=args.trust_request_chat_template,
)
state.vllm_config = vllm_config
# Disable stats logging — there is no engine to poll.