[Misc] Fix Current vLLM config is not set. warnings, assert to avoid issues in the future (#31747)

Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>

tests/kernels/quantization/test_fp8_quant_group.py

@@ -23,7 +23,12 @@ from vllm.utils.torch_utils import set_random_seed
 @pytest.mark.parametrize("use_ue8m0", [True, False])
 @torch.inference_mode()
 def test_quantfp8_group_functionality(
-    batch_size: int, hidden_dim: int, group_size: int, seed: int, use_ue8m0: bool
+    default_vllm_config,
+    batch_size: int,
+    hidden_dim: int,
+    group_size: int,
+    seed: int,
+    use_ue8m0: bool,
 ) -> None:
     """Test QuantFP8 group quantization with various configurations.
 
@@ -82,7 +87,9 @@ def test_quantfp8_group_functionality(
 @pytest.mark.parametrize("seed", [42])
 @pytest.mark.parametrize("use_ue8m0", [True, False])
 @torch.inference_mode()
-def test_quantfp8_group_multidimensional(seed: int, use_ue8m0: bool) -> None:
+def test_quantfp8_group_multidimensional(
+    default_vllm_config, seed: int, use_ue8m0: bool
+) -> None:
     set_random_seed(seed)
 
     group_size = 64
@@ -135,7 +142,7 @@ def test_quantfp8_group_multidimensional(seed: int, use_ue8m0: bool) -> None:
 
 @pytest.mark.parametrize("seed", [42])
 @torch.inference_mode()
-def test_quantfp8_group_edge_cases(seed: int) -> None:
+def test_quantfp8_group_edge_cases(default_vllm_config, seed: int) -> None:
     set_random_seed(seed)
 
     batch_size = 16

tests/kernels/quantization/test_int8_kernel.py

@@ -102,7 +102,7 @@ SEEDS = [0]
     itertools.product(M, N, K, E, TOP_KS, DTYPES, SEEDS),
 )
 @torch.inference_mode()
-def test_w8a8_fp8_fused_moe(M, N, K, E, topk, dtype, seed):
+def test_w8a8_fp8_fused_moe(default_vllm_config, M, N, K, E, topk, dtype, seed):
     torch.manual_seed(seed)
     # Initialize int8 quantization parameters
     factor_for_scale = 1e-2

tests/kernels/quantization/test_silu_mul_nvfp4_quant.py

@@ -31,6 +31,7 @@ BLOCK_SIZE = 16
 @pytest.mark.parametrize("shape", SHAPES)
 @torch.inference_mode()
 def test_silu_mul_nvfp4_quant(
+    default_vllm_config,
     dtype: torch.dtype,
     shape: tuple[int, int],
 ) -> None: