Fix per file ruff ignores related to line length (#26262)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

tests/kernels/attention/test_attention_selector.py

@@ -165,7 +165,7 @@ def test_env(
                             # FlashMLA only supports block_size == 64
                             pytest.skip("FlashMLA only supports block_size 64")
                         else:
-                            from vllm.v1.attention.backends.mla.flashmla import (  # noqa: E501
+                            from vllm.v1.attention.backends.mla.flashmla import (
                                 is_flashmla_supported,
                             )
 

tests/kernels/moe/modular_kernel_tools/common.py

@@ -331,7 +331,8 @@ class WeightTensors:
             in_dtype=config.dtype,
             quant_dtype=config.quant_dtype,
             block_shape=config.quant_block_shape,
-            per_out_ch_quant=config.is_per_act_token_quant,  # or config.is_per_out_ch_quant
+            # or config.is_per_out_ch_quant
+            per_out_ch_quant=config.is_per_act_token_quant,
         )
         return WeightTensors(
             w1=w1, w2=w2, w1_scale=w1_scale, w2_scale=w2_scale, w1_gs=w1_gs, w2_gs=w2_gs

tests/kernels/moe/modular_kernel_tools/make_feature_matrix.py

@@ -124,7 +124,7 @@ def make_feature_matrix(csv_file_path: str):
     results_df: Optional[pd.DataFrame] = None
     for m, k, n, e, topks, dtype, pf_type, experts_type, quant_config in tqdm(
         combinations
-    ):  # noqa: E501
+    ):
         config = Config(
             Ms=[m],
             K=k,

tests/kernels/moe/modular_kernel_tools/mk_objects.py

@@ -10,7 +10,7 @@ import vllm.model_executor.layers.fused_moe.modular_kernel as mk
 from vllm.model_executor.layers.fused_moe.batched_deep_gemm_moe import (
     BatchedDeepGemmExperts,
 )
-from vllm.model_executor.layers.fused_moe.batched_triton_or_deep_gemm_moe import (  # noqa: E501
+from vllm.model_executor.layers.fused_moe.batched_triton_or_deep_gemm_moe import (
     BatchedTritonOrDeepGemmExperts,
 )
 from vllm.model_executor.layers.fused_moe.config import (
@@ -196,10 +196,10 @@ register_experts(
 
 # Disable on blackwell for now
 if has_deep_ep() and not current_platform.has_device_capability(100):
-    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (
         DeepEPHTPrepareAndFinalize,
     )
-    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (
         DeepEPLLPrepareAndFinalize,
     )
 
@@ -233,7 +233,7 @@ if has_pplx():
     )
 
 if has_flashinfer_cutlass_fused_moe() and current_platform.has_device_capability(100):
-    from vllm.model_executor.layers.fused_moe.flashinfer_cutlass_moe import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.flashinfer_cutlass_moe import (
         FlashInferExperts,
     )
     from vllm.model_executor.layers.fused_moe.flashinfer_cutlass_prepare_finalize import (  # noqa: E501

tests/kernels/moe/parallel_utils.py

@@ -17,10 +17,10 @@ from typing_extensions import Concatenate, ParamSpec
 from vllm.utils import get_open_port, has_deep_ep
 
 if has_deep_ep():
-    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (
         DeepEPHTPrepareAndFinalize,
     )
-    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (
         DeepEPLLPrepareAndFinalize,
     )
 

tests/kernels/moe/test_deepep_deepgemm_moe.py

@@ -30,10 +30,10 @@ from .parallel_utils import ProcessGroupInfo, parallel_launch
 from .utils import make_test_weights
 
 if has_deep_ep():
-    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (
         DeepEPHTPrepareAndFinalize,
     )
-    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (
         DeepEPLLPrepareAndFinalize,
     )
 

tests/kernels/moe/test_deepep_moe.py

@@ -28,10 +28,10 @@ from ...utils import multi_gpu_test
 from .parallel_utils import ProcessGroupInfo, parallel_launch
 
 if has_deep_ep():
-    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ht_prepare_finalize import (
         DeepEPHTPrepareAndFinalize,
     )
-    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (  # noqa: E501
+    from vllm.model_executor.layers.fused_moe.deepep_ll_prepare_finalize import (
         DeepEPLLPrepareAndFinalize,
     )
 

tests/kernels/moe/test_modular_kernel_combinations.py

@@ -271,7 +271,7 @@ if __name__ == "__main__":
     parser = make_config_arg_parser(
         description=(
             "Run single prepare-finalize & fused-experts combination test"
-            "Example : python3 -m tests.kernels.moe.test_modular_kernel_combinations "  # noqa: E501
+            "Example : python3 -m tests.kernels.moe.test_modular_kernel_combinations "
             "--pf-type PplxPrepareAndFinalize --experts-type BatchedTritonExperts"
         )
     )

tests/kernels/moe/test_moe.py

@@ -483,8 +483,8 @@ def test_mixtral_moe(
     }
 
     if use_rocm_aiter:
-        # The values of rtol and atol are set based on the tests in ROCM AITER package. # noqa: E501
-        # https://github.com/ROCm/aiter/blob/dfed377f4be7da96ca2d75ac0761f569676f7240/op_tests/test_moe.py#L174  # noqa: E501
+        # The values of rtol and atol are set based on the tests in ROCM AITER package.
+        # https://github.com/ROCm/aiter/blob/dfed377f4be7da96ca2d75ac0761f569676f7240/op_tests/test_moe.py#L174
         torch.testing.assert_close(
             hf_states.flatten(0, 1), vllm_states, rtol=0.01, atol=100
         )

tests/kernels/moe/test_mxfp4_moe.py

@@ -10,11 +10,11 @@ import pytest
 import torch
 from packaging import version
 
-from vllm.model_executor.layers.quantization.quark.quark import (  # noqa: E501
+from vllm.model_executor.layers.quantization.quark.quark import (
     QuarkLinearMethod,
     QuarkW4A4MXFP4,
 )
-from vllm.model_executor.layers.quantization.quark.quark_moe import (  # noqa: E501
+from vllm.model_executor.layers.quantization.quark.quark_moe import (
     QuarkW4A4MXFp4MoEMethod,
 )
 from vllm.platforms import current_platform