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

[Bugfix] Fix auto dtype casting for BatchFeature (#19316)

Browse Source 
Signed-off-by: Isotr0py <2037008807@qq.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
This commit is contained in:
Isotr0py
2025-06-14 23:13:08 +08:00
committed by GitHub
parent 6fa718a460
commit 2db9044ab6
7 changed files with 85 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
tests/v1/engine/test_async_llm.py
View File

@@ -15,6 +15,7 @@ from vllm.engine.arg_utils import AsyncEngineArgs
from vllm.inputs import PromptType
from vllm.platforms import current_platform
from vllm.sampling_params import RequestOutputKind
from vllm.utils import set_default_torch_num_threads
from vllm.v1.engine.async_llm import AsyncLLM
from vllm.v1.metrics.loggers import LoggingStatLogger
@@ -107,7 +108,8 @@ async def test_load(
with monkeypatch.context() as m, ExitStack() as after:
m.setenv("VLLM_USE_V1", "1")
engine = AsyncLLM.from_engine_args(engine_args)
with set_default_torch_num_threads(1):
engine = AsyncLLM.from_engine_args(engine_args)
after.callback(engine.shutdown)
NUM_REQUESTS = 100
@@ -154,7 +156,8 @@ async def test_abort(
with monkeypatch.context() as m, ExitStack() as after:
m.setenv("VLLM_USE_V1", "1")
engine = AsyncLLM.from_engine_args(engine_args)
with set_default_torch_num_threads(1):
engine = AsyncLLM.from_engine_args(engine_args)
after.callback(engine.shutdown)
NUM_REQUESTS = 100
@@ -226,7 +229,8 @@ async def test_finished_flag(
with monkeypatch.context() as m, ExitStack() as after:
m.setenv("VLLM_USE_V1", "1")
engine = AsyncLLM.from_engine_args(engine_args)
with set_default_torch_num_threads(1):
engine = AsyncLLM.from_engine_args(engine_args)
after.callback(engine.shutdown)
sampling_params = SamplingParams(
@@ -260,7 +264,8 @@ async def test_mid_stream_cancellation(monkeypatch: pytest.MonkeyPatch,
with monkeypatch.context() as m, ExitStack() as after:
m.setenv("VLLM_USE_V1", "1")
engine = AsyncLLM.from_engine_args(engine_args)
with set_default_torch_num_threads(1):
engine = AsyncLLM.from_engine_args(engine_args)
after.callback(engine.shutdown)
NUM_REQUESTS = 100
@@ -322,10 +327,11 @@ async def test_customize_loggers(monkeypatch):
with monkeypatch.context() as m, ExitStack() as after:
m.setenv("VLLM_USE_V1", "1")
engine = AsyncLLM.from_engine_args(
TEXT_ENGINE_ARGS,
stat_loggers=[MockLoggingStatLogger],
)
with set_default_torch_num_threads(1):
engine = AsyncLLM.from_engine_args(
TEXT_ENGINE_ARGS,
stat_loggers=[MockLoggingStatLogger],
)
after.callback(engine.shutdown)
await engine.do_log_stats()
@@ -340,7 +346,8 @@ async def test_dp_rank_argument(monkeypatch: pytest.MonkeyPatch):
with monkeypatch.context() as m, ExitStack() as after:
m.setenv("VLLM_USE_V1", "1")
engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
with set_default_torch_num_threads(1):
engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
after.callback(engine.shutdown)
sampling_params = SamplingParams(max_tokens=100,