A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 13, 08:15 UTC
Fix io processor pooling #28273 (#28484)
Signed-off-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
baonudesifeizhai committed
Nov 13, 2025 at 06:34 UTC
c428e8d80b2bc17b0a306d1e80c8e4567b9dd9f4
1 file changed
+5
-1
vllm/entrypoints/openai/serving_pooling.py
+5
-1
@@ -4,7 +4,7 @@
4
import asyncio
5
import json
6
import time
7
-from collections.abc import AsyncGenerator
7
+from collections.abc import AsyncGenerator, Sequence
8
from typing import Final, cast
9
10
import jinja2
@@ -122,6 +122,10 @@ class OpenAIServingPooling(OpenAIServing):
122
engine_prompts = await self.io_processor.pre_process_async(
123
prompt=validated_prompt, request_id=request_id
124
)
125
+ if not isinstance(engine_prompts, Sequence) or isinstance(
126
+ engine_prompts, (str, bytes, bytearray)
127
+ ):
128
+ engine_prompts = [engine_prompts]
129
130
elif isinstance(request, PoolingChatRequest):
131
error_check_ret = self._validate_chat_template(