diff --git a/tests/v1/tpu/test_mha_attn.py b/tests/v1/tpu/test_mha_attn.py index 55fee4ee1ad4..9d690851b70e 100644 --- a/tests/v1/tpu/test_mha_attn.py +++ b/tests/v1/tpu/test_mha_attn.py @@ -12,17 +12,10 @@ import torch_xla.core import torch_xla.core.xla_model -from vllm import envs from vllm.attention.layer import MultiHeadAttention from vllm.attention.selector import _cached_get_attn_backend from vllm.platforms import current_platform -if not envs.VLLM_USE_V1: - pytest.skip( - "Skipping V1 tests. Rerun with `VLLM_USE_V1=1` to test.", - allow_module_level=True, - ) - @pytest.fixture(autouse=True) def clear_cache(): diff --git a/tests/v1/tpu/test_multimodal.py b/tests/v1/tpu/test_multimodal.py index a61773a4f611..bcc2993028dd 100644 --- a/tests/v1/tpu/test_multimodal.py +++ b/tests/v1/tpu/test_multimodal.py @@ -4,19 +4,12 @@ import openai import pytest -from vllm import envs from vllm.multimodal.utils import encode_image_base64, fetch_image from vllm.platforms import current_platform from ...entrypoints.openai.test_vision import TEST_IMAGE_URLS from ...utils import RemoteOpenAIServer -if not envs.VLLM_USE_V1: - pytest.skip( - "Skipping V1 tests. Rerun with `VLLM_USE_V1=1` to test.", - allow_module_level=True, - ) - @pytest.fixture(scope="session") def base64_encoded_image() -> dict[str, str]: diff --git a/tests/v1/tpu/test_sampler.py b/tests/v1/tpu/test_sampler.py index 198bb1e16ed9..fa950e5f7f85 100644 --- a/tests/v1/tpu/test_sampler.py +++ b/tests/v1/tpu/test_sampler.py @@ -4,16 +4,10 @@ import pytest -from vllm import LLM, envs +from vllm import LLM from vllm.platforms import current_platform from vllm.sampling_params import SamplingParams -if not envs.VLLM_USE_V1: - pytest.skip( - "Skipping V1 tests. Rerun with `VLLM_USE_V1=1` to test.", - allow_module_level=True, - ) - @pytest.mark.parametrize("model_name", ["Qwen/Qwen2.5-1.5B-Instruct"]) @pytest.mark.skipif(not current_platform.is_tpu(),