xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 02:55:40 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix]: Fix the incompatibility issue with stream when Thinking is disabled (#19135)

Browse Source 
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
This commit is contained in:
Chauncey 2025-06-05 14:24:24 +08:00 committed by GitHub
parent af7fc84fd2
commit 8fc57501d3
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 98 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

78
tests/entrypoints/openai/test_completion_with_function_calling.py
View File

@ -1,6 +1,8 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
from typing import NamedTuple
import openai # use the official client for correctness check
import pytest
import pytest_asyncio
@ -22,7 +24,9 @@ def server(): # noqa: F811
"--guided-decoding-backend",
"xgrammar",
"--tool-call-parser",
"hermes"
"hermes",
"--reasoning-parser",
"qwen3",
]
with RemoteOpenAIServer(MODEL_NAME, args) as remote_server:
@ -35,9 +39,53 @@ async def client(server):
yield async_client
class TestCase(NamedTuple):
model_name: str
stream: bool
tool_choice: str
enable_thinking: bool
@pytest.mark.asyncio
@pytest.mark.parametrize("model_name", [MODEL_NAME])
async def test_required_tool_use(client: openai.AsyncOpenAI, model_name: str):
@pytest.mark.parametrize(
"test_case",
[
TestCase(model_name=MODEL_NAME,
stream=True,
tool_choice="auto",
enable_thinking=False),
TestCase(model_name=MODEL_NAME,
stream=False,
tool_choice="auto",
enable_thinking=False),
TestCase(model_name=MODEL_NAME,
stream=True,
tool_choice="required",
enable_thinking=False),
TestCase(model_name=MODEL_NAME,
stream=False,
tool_choice="required",
enable_thinking=False),
TestCase(model_name=MODEL_NAME,
stream=True,
tool_choice="auto",
enable_thinking=True),
TestCase(model_name=MODEL_NAME,
stream=False,
tool_choice="auto",
enable_thinking=True),
TestCase(model_name=MODEL_NAME,
stream=True,
tool_choice="required",
enable_thinking=True),
TestCase(model_name=MODEL_NAME,
stream=False,
tool_choice="required",
enable_thinking=True),
],
)
async def test_function_tool_use(client: openai.AsyncOpenAI,
test_case: TestCase):
tools = [
{
"type": "function",
@ -126,26 +174,34 @@ async def test_required_tool_use(client: openai.AsyncOpenAI, model_name: str):
"forecast for the next 5 days, in fahrenheit?",
},
]
if not test_case.stream:
# Non-streaming test
chat_completion = await client.chat.completions.create(
messages=messages,
model=model_name,
model=test_case.model_name,
tools=tools,
tool_choice="required",
)
tool_choice=test_case.tool_choice,
extra_body={
"chat_template_kwargs": {
"enable_thinking": test_case.enable_thinking
}
})
assert chat_completion.choices[0].message.tool_calls is not None
assert len(chat_completion.choices[0].message.tool_calls) > 0
else:
# Streaming test
stream = await client.chat.completions.create(
messages=messages,
model=model_name,
model=test_case.model_name,
tools=tools,
tool_choice="required",
tool_choice=test_case.tool_choice,
stream=True,
)
extra_body={
"chat_template_kwargs": {
"enable_thinking": test_case.enable_thinking
}
})
output = []
async for chunk in stream:

16
vllm/entrypoints/openai/serving_chat.py
View File

@ -689,7 +689,21 @@ class OpenAIServingChat(OpenAIServing):
current_token_ids,
output.token_ids,
))
# When encountering think end id in prompt_token_ids
# i.e {"enable_thinking": False},
# set reasoning status to end.
# Remove the text and token ids related
# to 'reasoning_content'.
if res.prompt_token_ids and \
reasoning_parser.is_reasoning_end(
list(res.prompt_token_ids)):
reasoning_end_arr[i] = True
current_token_ids = list(output.token_ids)
if delta_message and delta_message.content:
current_text = delta_message.content
delta_message.content = None
else:
current_text = ""
# When encountering think end id in delta_token_ids,
# set reasoning status to end.
# Remove the text and token ids related