From 311b3513af33bc29b4acb2fde2e9313e5e9966a0 Mon Sep 17 00:00:00 2001 From: Andreas Karatzas Date: Sun, 16 Aug 2026 23:08:35 -0500 Subject: [PATCH] [ROCm][CI] Avoid forcing FlashAttention in the ColPali pooling test (#52565) Signed-off-by: Andreas Karatzas --- tests/models/multimodal/pooling/test_colpali.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/tests/models/multimodal/pooling/test_colpali.py b/tests/models/multimodal/pooling/test_colpali.py index 41e6393dc7..86e595824d 100644 --- a/tests/models/multimodal/pooling/test_colpali.py +++ b/tests/models/multimodal/pooling/test_colpali.py @@ -19,6 +19,7 @@ from vllm.entrypoints.chat_utils import ( ChatCompletionContentPartTextParam, ) from vllm.entrypoints.pooling.scoring.typing import ScoreMultiModalParam +from vllm.platforms import current_platform from ....conftest import VllmRunner @@ -215,6 +216,7 @@ def _run_multimodal_text_query_image_docs_test( _make_image_mm_param(red_image), _make_image_mm_param(blue_image), ] + attention_backend = "FLASH_ATTN" if current_platform.is_cuda() else None with vllm_runner( model, @@ -223,7 +225,7 @@ def _run_multimodal_text_query_image_docs_test( max_model_len=4096, enforce_eager=True, gpu_memory_utilization=GPU_MEMORY_UTILIZATION, - attention_backend="FLASH_ATTN", + attention_backend=attention_backend, kernel_config={"enable_flashinfer_autotune": False}, ) as vllm_model: assert vllm_model.llm.llm_engine.vllm_config.use_v2_model_runner