From 0291fbf65cec9082dba4358f33f1d89ef595172c Mon Sep 17 00:00:00 2001 From: Zhewen Li Date: Tue, 28 Oct 2025 01:58:11 -0700 Subject: [PATCH] [CI/Build] Fix amd model executor test (#27612) Signed-off-by: zhewenli --- .buildkite/test-amd.yaml | 2 +- .../fastsafetensors_loader/test_fastsafetensors_loader.py | 6 ++++++ .../fastsafetensors_loader/test_weight_utils.py | 5 +++++ 3 files changed, 12 insertions(+), 1 deletion(-) diff --git a/.buildkite/test-amd.yaml b/.buildkite/test-amd.yaml index ad4030e1208f..524d2e121a10 100644 --- a/.buildkite/test-amd.yaml +++ b/.buildkite/test-amd.yaml @@ -561,7 +561,7 @@ steps: - label: Model Executor Test # 23min timeout_in_minutes: 35 - mirror_hardwares: [amdexperimental] + mirror_hardwares: [amdexperimental, amdproduction] agent_pool: mi325_1 # grade: Blocking source_file_dependencies: diff --git a/tests/model_executor/model_loader/fastsafetensors_loader/test_fastsafetensors_loader.py b/tests/model_executor/model_loader/fastsafetensors_loader/test_fastsafetensors_loader.py index afd411ff4874..f154df6dfc23 100644 --- a/tests/model_executor/model_loader/fastsafetensors_loader/test_fastsafetensors_loader.py +++ b/tests/model_executor/model_loader/fastsafetensors_loader/test_fastsafetensors_loader.py @@ -1,7 +1,10 @@ # SPDX-License-Identifier: Apache-2.0 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project +import pytest + from vllm import SamplingParams +from vllm.platforms import current_platform test_model = "openai-community/gpt2" @@ -15,6 +18,9 @@ prompts = [ sampling_params = SamplingParams(temperature=0.8, top_p=0.95, seed=0) +@pytest.mark.skipif( + not current_platform.is_cuda(), reason="fastsafetensors requires CUDA/NVIDIA GPUs" +) def test_model_loader_download_files(vllm_runner): with vllm_runner(test_model, load_format="fastsafetensors") as llm: deserialized_outputs = llm.generate(prompts, sampling_params) diff --git a/tests/model_executor/model_loader/fastsafetensors_loader/test_weight_utils.py b/tests/model_executor/model_loader/fastsafetensors_loader/test_weight_utils.py index cc899b77b5e9..bd216f0e41a4 100644 --- a/tests/model_executor/model_loader/fastsafetensors_loader/test_weight_utils.py +++ b/tests/model_executor/model_loader/fastsafetensors_loader/test_weight_utils.py @@ -5,6 +5,7 @@ import glob import tempfile import huggingface_hub.constants +import pytest import torch from vllm.model_executor.model_loader.weight_utils import ( @@ -12,8 +13,12 @@ from vllm.model_executor.model_loader.weight_utils import ( fastsafetensors_weights_iterator, safetensors_weights_iterator, ) +from vllm.platforms import current_platform +@pytest.mark.skipif( + not current_platform.is_cuda(), reason="fastsafetensors requires CUDA/NVIDIA GPUs" +) def test_fastsafetensors_model_loader(): with tempfile.TemporaryDirectory() as tmpdir: huggingface_hub.constants.HF_HUB_OFFLINE = False