[Frontend][4/n] Improve pooling entrypoints | pooling. (#39153)
Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io>
This commit is contained in:
@@ -67,20 +67,18 @@ def init_pooling_state(
|
||||
from vllm.entrypoints.chat_utils import load_chat_template
|
||||
from vllm.entrypoints.pooling.classify.serving import ServingClassification
|
||||
from vllm.entrypoints.pooling.embed.serving import ServingEmbedding
|
||||
from vllm.entrypoints.pooling.pooling.serving import OpenAIServingPooling
|
||||
from vllm.entrypoints.pooling.pooling.serving import ServingPooling
|
||||
from vllm.entrypoints.pooling.scoring.serving import ServingScores
|
||||
from vllm.tasks import POOLING_TASKS
|
||||
|
||||
model_config = engine_client.model_config
|
||||
|
||||
resolved_chat_template = load_chat_template(args.chat_template)
|
||||
|
||||
state.serving_pooling = (
|
||||
(
|
||||
OpenAIServingPooling(
|
||||
ServingPooling(
|
||||
engine_client,
|
||||
state.openai_serving_models,
|
||||
state.openai_serving_render,
|
||||
supported_tasks=supported_tasks,
|
||||
request_logger=request_logger,
|
||||
chat_template=resolved_chat_template,
|
||||
|
||||
Reference in New Issue
Block a user