From a4123a53ec2a2d8a626614ac118b8e2b6390f712 Mon Sep 17 00:00:00 2001 From: tzhouam Date: Sun, 4 Oct 2026 03:26:52 +0800 Subject: [PATCH] [Test] Fit Qwen2.5 online serving placement to its two-card allocation Signed-off-by: tzhouam --- tests/e2e/online_serving/test_qwen2_5_omni_expansion.py | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/tests/e2e/online_serving/test_qwen2_5_omni_expansion.py b/tests/e2e/online_serving/test_qwen2_5_omni_expansion.py index 389b0b6ba..9c9bbc0b3 100644 --- a/tests/e2e/online_serving/test_qwen2_5_omni_expansion.py +++ b/tests/e2e/online_serving/test_qwen2_5_omni_expansion.py @@ -21,7 +21,13 @@ models = ["Qwen/Qwen2.5-Omni-7B"] # Single CI deploy YAML; rocm/xpu deltas are picked automatically via the # platforms: section in vllm_omni/deploy/ci/qwen2_5_omni.yaml. -stage_configs = [modify_stage_config(get_deploy_config_path("ci/qwen2_5_omni.yaml"))] +stage_configs = [ + modify_stage_config( + get_deploy_config_path("ci/qwen2_5_omni.yaml"), + # This suite requests two cards; Code2Wav shares the Talker's card. + updates={"stages": {2: {"devices": "1"}}}, + ) +] # Create parameter combinations for model and stage config test_params = [