A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[BugFix] Update KV block hash type from BlockHash to ExternalBlockHash in kv_events_subscriber - #26264 (#26265)
Signed-off-by: atalhens <sneh.lata@nutanix.com>
Snehlata committed
Oct 7, 2025 at 14:12 UTC
46b0779996dd689ca7e5e292a34c2909967915c1
1 file changed
+4
-4
examples/online_serving/kv_events_subscriber.py
+4
-4
@@ -6,7 +6,7 @@ import msgspec
6
import zmq
7
from msgspec.msgpack import Decoder
8
9
-from vllm.v1.core.kv_cache_utils import BlockHash
9
+from vllm.v1.core.kv_cache_utils import ExternalBlockHash
10
11
12
#
@@ -24,8 +24,8 @@ class KVCacheEvent(
24
25
26
class BlockStored(KVCacheEvent):
27
- block_hashes: list[BlockHash]
28
- parent_block_hash: Optional[BlockHash]
27
+ block_hashes: list[ExternalBlockHash]
28
+ parent_block_hash: Optional[ExternalBlockHash]
29
token_ids: list[int]
30
block_size: int
31
lora_id: Optional[int]
@@ -33,7 +33,7 @@ class BlockStored(KVCacheEvent):
33
34
35
class BlockRemoved(KVCacheEvent):
36
- block_hashes: list[BlockHash]
36
+ block_hashes: list[ExternalBlockHash]
37
medium: Optional[str]
38
39