[Typing] Fix Sequence type GenericAlias only available after Python 3.9. (#4092)

vllm/core/block_manager_v1.py

@@ -1,9 +1,10 @@
 """A block manager that manages token blocks."""
 from abc import ABC, abstractmethod
-from collections.abc import Sequence as GenericSequence
 from itertools import count, takewhile
 from os.path import commonprefix
-from typing import Dict, List, Optional, Set
+from typing import Dict, List, Optional
+from typing import Sequence as GenericSequence
+from typing import Set
 
 from vllm.block import BlockTable, PhysicalTokenBlock
 from vllm.core.evictor import EvictionPolicy, Evictor, make_evictor

vllm/core/block_manager_v2.py

@@ -1,6 +1,6 @@
 """A block manager that manages token blocks."""
-from collections.abc import Sequence as GenericSequence
 from typing import Dict, List, Optional
+from typing import Sequence as GenericSequence
 
 from vllm.core.block.block_table import BlockTable
 from vllm.core.block.cpu_gpu_block_allocator import CpuGpuBlockAllocator

vllm/core/interfaces.py

@@ -1,7 +1,7 @@
 import enum
 from abc import ABC, abstractmethod
-from collections.abc import Sequence as GenericSequence
 from typing import Dict, List
+from typing import Sequence as GenericSequence
 
 from vllm.sequence import Sequence, SequenceGroup
 

vllm/utils.py

@@ -6,11 +6,12 @@ import socket
 import subprocess
 import uuid
 import warnings
-from collections import OrderedDict, defaultdict
+from collections import defaultdict
 from functools import lru_cache, partial
 from platform import uname
 from typing import (Any, AsyncIterator, Awaitable, Callable, Dict, Generic,
-                    Hashable, List, Optional, Tuple, TypeVar, Union)
+                    Hashable, List, Optional, OrderedDict, Tuple, TypeVar,
+                    Union)
 
 import psutil
 import torch
@@ -51,7 +52,7 @@ class Counter:
 class LRUCache(Generic[T]):
 
     def __init__(self, capacity: int):
-        self.cache = OrderedDict[Hashable, T]()
+        self.cache: OrderedDict[Hashable, T] = OrderedDict()
         self.capacity = capacity
 
     def __contains__(self, key: Hashable) -> bool: