[CI] Stabilize memory-sensitive compile and structured output tests (#49749)
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com> Co-authored-by: OpenAI Codex <noreply@openai.com>
This commit is contained in:
co-authored by
OpenAI Codex
parent
972848f276
commit
9e6746b3c7
@@ -64,6 +64,8 @@ class TestSetting:
|
||||
"bfloat16",
|
||||
"--max-model-len",
|
||||
"2048",
|
||||
"--gpu-memory-utilization",
|
||||
"0.98",
|
||||
],
|
||||
pp_size=1,
|
||||
tp_size=1,
|
||||
|
||||
@@ -212,6 +212,7 @@ class CarDescription(BaseModel):
|
||||
PARAMS_MODELS_BACKENDS_TOKENIZER_MODE,
|
||||
)
|
||||
def test_structured_output(
|
||||
request: pytest.FixtureRequest,
|
||||
backend: str,
|
||||
tokenizer_mode: str,
|
||||
model_name: str,
|
||||
@@ -242,6 +243,7 @@ def test_structured_output(
|
||||
speculative_config=speculative_config,
|
||||
**platform_args,
|
||||
)
|
||||
request.addfinalizer(llm.llm_engine.engine_core.shutdown)
|
||||
|
||||
#
|
||||
# Test 1: Generate JSON output based on a provided schema
|
||||
|
||||
Reference in New Issue
Block a user