From 773d7073aea7de91710dad24928f46497017a8f6 Mon Sep 17 00:00:00 2001 From: Micah Williamson Date: Wed, 14 Jan 2026 22:53:43 -0600 Subject: [PATCH] [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 --- tests/v1/entrypoints/llm/test_struct_output_generate.py | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/tests/v1/entrypoints/llm/test_struct_output_generate.py b/tests/v1/entrypoints/llm/test_struct_output_generate.py index d819a92e5d..c6c9c0ce40 100644 --- a/tests/v1/entrypoints/llm/test_struct_output_generate.py +++ b/tests/v1/entrypoints/llm/test_struct_output_generate.py @@ -87,6 +87,10 @@ PARAMS_MODELS_TOKENIZER_MODE = [ ("Qwen/Qwen2.5-1.5B-Instruct", "auto"), ] +platform_args = {} +if current_platform.is_rocm(): + platform_args["async_scheduling"] = False + class CarType(str, Enum): sedan = "sedan" @@ -134,6 +138,7 @@ def test_structured_output( load_format="auto" if not model_name.startswith("mistralai/") else "hf", config_format="auto" if not model_name.startswith("mistralai/") else "hf", speculative_config=speculative_config, + **platform_args, ) #