[Frontend][4/n] Improve pooling entrypoints | pooling. (#39153)

Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io>
This commit is contained in:
wang.yuqi
2026-04-09 10:09:45 +00:00
committed by GitHub
parent b6c9be509e
commit 66c079ae83
43 changed files with 554 additions and 733 deletions
+10 -18
View File
@@ -8,8 +8,6 @@ from typing import Literal, TypeAlias, cast
from fastapi.responses import JSONResponse, Response, StreamingResponse
from typing_extensions import assert_never
from vllm.config import ModelConfig
from vllm.entrypoints.chat_utils import ChatTemplateConfig
from vllm.entrypoints.openai.engine.protocol import UsageInfo
from vllm.entrypoints.pooling.base.serving import PoolingServing
from vllm.entrypoints.pooling.embed.io_processor import EmbedIOProcessor
@@ -33,12 +31,10 @@ from vllm.entrypoints.pooling.utils import (
)
from vllm.logger import init_logger
from vllm.outputs import PoolingRequestOutput
from vllm.renderers import BaseRenderer
from vllm.utils.serial_utils import EmbedDType, Endianness
logger = init_logger(__name__)
JSONResponseCLS = get_json_response_cls()
EmbeddingServeContext: TypeAlias = PoolingServeContext[EmbeddingRequest]
@@ -49,17 +45,13 @@ class ServingEmbedding(PoolingServing):
request_id_prefix = "embd"
io_processor: EmbedIOProcessor
def init_io_processor(
self,
model_config: ModelConfig,
renderer: BaseRenderer,
chat_template_config: ChatTemplateConfig,
) -> EmbedIOProcessor:
return EmbedIOProcessor(
model_config=model_config,
renderer=renderer,
chat_template_config=chat_template_config,
)
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.json_response_cls = get_json_response_cls()
def init_io_processor(self, *args, **kwargs) -> EmbedIOProcessor:
return EmbedIOProcessor(*args, **kwargs)
async def _build_response(
self,
@@ -149,7 +141,7 @@ class ServingEmbedding(PoolingServing):
data=items,
usage=usage,
)
return JSONResponseCLS(content=response.model_dump())
return self.json_response_cls(content=response.model_dump())
def _openai_bytes_response(
self,
@@ -190,8 +182,8 @@ class ServingEmbedding(PoolingServing):
media_type=response.media_type,
)
@staticmethod
def _build_cohere_response_from_ctx(
self,
ctx: PoolingServeContext,
) -> JSONResponse:
request = ctx.request
@@ -218,4 +210,4 @@ class ServingEmbedding(PoolingServing):
),
),
)
return JSONResponse(content=response.model_dump(exclude_none=True))
return self.json_response_cls(content=response.model_dump(exclude_none=True))