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

Convert formatting to use ruff instead of yapf + isort (#26247)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-10-05 15:06:22 +01:00
committed by GitHub
parent 17edd8a807
commit d6953beb91
1508 changed files with 115244 additions and 94146 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
tests/entrypoints/openai/test_sleep.py
View File

@@ -20,14 +20,12 @@ def test_sleep_mode():
"--enable-sleep-mode",
]
with RemoteOpenAIServer(MODEL_NAME,
args,
env_dict={
"VLLM_SERVER_DEV_MODE": "1",
"CUDA_VISIBLE_DEVICES": "0"
}) as remote_server:
response = requests.post(remote_server.url_for("sleep"),
params={"level": "1"})
with RemoteOpenAIServer(
MODEL_NAME,
args,
env_dict={"VLLM_SERVER_DEV_MODE": "1", "CUDA_VISIBLE_DEVICES": "0"},
) as remote_server:
response = requests.post(remote_server.url_for("sleep"), params={"level": "1"})
assert response.status_code == 200
response = requests.get(remote_server.url_for("is_sleeping"))
assert response.status_code == 200
@@ -40,12 +38,12 @@ def test_sleep_mode():
assert response.json().get("is_sleeping") is False
# test wake up with tags
response = requests.post(remote_server.url_for("sleep"),
params={"level": "1"})
response = requests.post(remote_server.url_for("sleep"), params={"level": "1"})
assert response.status_code == 200
response = requests.post(remote_server.url_for("wake_up"),
params={"tags": ["weights"]})
response = requests.post(
remote_server.url_for("wake_up"), params={"tags": ["weights"]}
)
assert response.status_code == 200
# is sleeping should be false after waking up any part of the engine
@@ -53,8 +51,9 @@ def test_sleep_mode():
assert response.status_code == 200
assert response.json().get("is_sleeping") is True
response = requests.post(remote_server.url_for("wake_up"),
params={"tags": ["kv_cache"]})
response = requests.post(
remote_server.url_for("wake_up"), params={"tags": ["kv_cache"]}
)
assert response.status_code == 200
response = requests.get(remote_server.url_for("is_sleeping"))