Update Optional[x] -> x | None and Union[x, y] to x | y (#26633)

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

vllm/v1/attention/backends/triton_attn.py

@@ -3,7 +3,7 @@
 """High-Performance Triton-only Attention layer."""
 
 from dataclasses import dataclass
-from typing import ClassVar, Optional, Union
+from typing import ClassVar
 
 import torch
 
@@ -61,13 +61,13 @@ class TritonAttentionMetadata:
     # For cascade attention.
     use_cascade: bool
     common_prefix_len: int
-    cu_prefix_query_lens: Optional[torch.Tensor]
-    prefix_kv_lens: Optional[torch.Tensor]
-    suffix_kv_lens: Optional[torch.Tensor]
+    cu_prefix_query_lens: torch.Tensor | None
+    prefix_kv_lens: torch.Tensor | None
+    suffix_kv_lens: torch.Tensor | None
 
     # Optional aot scheduling
-    scheduler_metadata: Optional[torch.Tensor] = None
-    prefix_scheduler_metadata: Optional[torch.Tensor] = None
+    scheduler_metadata: torch.Tensor | None = None
+    prefix_scheduler_metadata: torch.Tensor | None = None
 
 
 class TritonAttentionMetadataBuilder(AttentionMetadataBuilder[TritonAttentionMetadata]):
@@ -159,7 +159,7 @@ class TritonAttentionBackend(AttentionBackend):
         return [torch.float16, torch.bfloat16, torch.float32]
 
     @staticmethod
-    def get_supported_kernel_block_size() -> list[Union[int, MultipleOf]]:
+    def get_supported_kernel_block_size() -> list[int | MultipleOf]:
         return [MultipleOf(16)]
 
     @classmethod
@@ -216,13 +216,13 @@ class TritonAttentionImpl(AttentionImpl):
         head_size: int,
         scale: float,
         num_kv_heads: int,
-        alibi_slopes: Optional[list[float]],
-        sliding_window: Optional[int],
+        alibi_slopes: list[float] | None,
+        sliding_window: int | None,
         kv_cache_dtype: str,
-        logits_soft_cap: Optional[float] = None,
+        logits_soft_cap: float | None = None,
         attn_type: AttentionType = AttentionType.DECODER,
-        kv_sharing_target_layer_name: Optional[int] = None,
-        sinks: Optional[torch.Tensor] = None,
+        kv_sharing_target_layer_name: int | None = None,
+        sinks: torch.Tensor | None = None,
     ) -> None:
         self.num_heads = num_heads
         self.head_size = head_size
@@ -272,9 +272,9 @@ class TritonAttentionImpl(AttentionImpl):
         value: torch.Tensor,
         kv_cache: torch.Tensor,
         attn_metadata: TritonAttentionMetadata,
-        output: Optional[torch.Tensor] = None,
-        output_scale: Optional[torch.Tensor] = None,
-        output_block_scale: Optional[torch.Tensor] = None,
+        output: torch.Tensor | None = None,
+        output_scale: torch.Tensor | None = None,
+        output_block_scale: torch.Tensor | None = None,
     ) -> torch.Tensor:
         """Forward pass with Paged Attention impl. in Triton.