biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V0 Deprecation] Remove AsyncLLMEngine (#25025)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon
2025-09-18 11:07:42 -07:00
committed by GitHub
parent 505805b645
commit e19bce40a1
11 changed files with 76 additions and 1967 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
tests/entrypoints/openai/test_chat.py
View File

@@ -28,11 +28,9 @@ def monkeypatch_module():
mpatch.undo()
@pytest.fixture(scope="module", params=[False, True])
def server(request, monkeypatch_module, zephyr_lora_files): #noqa: F811
use_v1 = request.param
monkeypatch_module.setenv('VLLM_USE_V1', '1' if use_v1 else '0')
@pytest.fixture(scope="module")
def server(monkeypatch_module, zephyr_lora_files): #noqa: F811
monkeypatch_module.setenv('VLLM_USE_V1', '1')
args = [
# use half precision for speed and memory savings in CI environment
@@ -57,13 +55,6 @@ def server(request, monkeypatch_module, zephyr_lora_files): #noqa: F811
yield remote_server
@pytest.fixture
def is_v1_server(server):
import os
assert os.environ['VLLM_USE_V1'] in ['0', '1']
return os.environ['VLLM_USE_V1'] == '1'
@pytest_asyncio.fixture
async def client(server):
async with server.get_async_client() as async_client:
@@ -481,10 +472,9 @@ async def test_chat_completion_stream_options(client: openai.AsyncOpenAI,
@pytest.mark.asyncio
async def test_structured_outputs_choice_chat(
client: openai.AsyncOpenAI, sample_structured_outputs_choices,
is_v1_server: bool):
if not is_v1_server:
pytest.skip("Structured outputs is only supported in v1 engine")
client: openai.AsyncOpenAI,
sample_structured_outputs_choices,
):
messages = [{
"role": "system",
"content": "you are a helpful assistant"
@@ -522,12 +512,10 @@ async def test_structured_outputs_choice_chat(
@pytest.mark.asyncio
async def test_structured_outputs_json_chat(client: openai.AsyncOpenAI,
sample_json_schema,
is_v1_server: bool):
if not is_v1_server:
pytest.skip("Structured outputs is only supported in v1 engine")
async def test_structured_outputs_json_chat(
client: openai.AsyncOpenAI,
sample_json_schema,
):
messages = [{
"role": "system",
"content": "you are a helpful assistant"
@@ -569,10 +557,10 @@ async def test_structured_outputs_json_chat(client: openai.AsyncOpenAI,
@pytest.mark.asyncio
async def test_structured_outputs_regex_chat(client: openai.AsyncOpenAI,
sample_regex, is_v1_server: bool):
if not is_v1_server:
pytest.skip("Structured outputs is only supported in v1 engine")
async def test_structured_outputs_regex_chat(
client: openai.AsyncOpenAI,
sample_regex,
):
messages = [{
"role": "system",
@@ -660,10 +648,10 @@ async def test_structured_outputs_choice_chat_logprobs(
@pytest.mark.asyncio
async def test_named_tool_use(client: openai.AsyncOpenAI, sample_json_schema,
is_v1_server: bool):
if not is_v1_server:
pytest.skip("Tool use is only supported in v1 engine")
async def test_named_tool_use(
client: openai.AsyncOpenAI,
sample_json_schema,
):
messages = [{
"role": "system",
"content": "you are a helpful assistant"
@@ -821,11 +809,7 @@ async def test_response_format_json_object(client: openai.AsyncOpenAI):
@pytest.mark.asyncio
async def test_response_format_json_schema(client: openai.AsyncOpenAI,
is_v1_server: bool):
if not is_v1_server:
pytest.skip(
"JSON schema response format is only supported in v1 engine")
async def test_response_format_json_schema(client: openai.AsyncOpenAI):
prompt = 'what is 1+1? The format is "result": 2'
# Check that this prompt cannot lead to a valid JSON without json_schema
for _ in range(2):