A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 10, 04:00 UTC
[ROCm][CI] Disable async scheduling on ROCm for test_structured_output[meta-llama/Meta-Llama-3.1-8B-Instruct-xgrammar-auto-speculative_config9] (#32355)
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
Micah Williamson committed
Jan 14, 2026 at 22:53 UTC
773d7073aea7de91710dad24928f46497017a8f6
1 file changed
+5
tests/v1/entrypoints/llm/test_struct_output_generate.py
+5
@@ -87,6 +87,10 @@ PARAMS_MODELS_TOKENIZER_MODE = [
87
("Qwen/Qwen2.5-1.5B-Instruct", "auto"),
88
]
89
90
+platform_args = {}
91
+if current_platform.is_rocm():
92
+ platform_args["async_scheduling"] = False
93
+
94
95
class CarType(str, Enum):
96
sedan = "sedan"
@@ -134,6 +138,7 @@ def test_structured_output(
138
load_format="auto" if not model_name.startswith("mistralai/") else "hf",
139
config_format="auto" if not model_name.startswith("mistralai/") else "hf",
140
speculative_config=speculative_config,
141
+ **platform_args,
142
)
143
144
#