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, 22:35 UTC
[Bugfix] Fix comment typo of get_num_common_prefix_blocks() (#21827)
Signed-off-by: MingzhenHan <hanmingzhen2002@outlook.com>
MingzhenHan committed
Jul 30, 2025 at 11:35 UTC
b7b23da4d25add19411821fa5f784529d4de8732
2 files changed
+3
-3
vllm/v1/core/kv_cache_coordinator.py
+2
-2
@@ -130,10 +130,10 @@ class KVCacheCoordinator(ABC):
130
131
Args:
132
request_id: The request ID.
133
- block_hashes: The block hashes of the request.
133
+ num_running_requests: The number of requests in the RUNNING state.
134
135
Returns:
136
- The number of common prefix blocks.
136
+ list[int]: The number of common prefix blocks.
137
"""
138
num_blocks_per_group = [
139
manager.get_num_common_prefix_blocks(request_id,
vllm/v1/core/single_type_kv_cache_manager.py
+1
-1
@@ -181,7 +181,7 @@ class SingleTypeKVCacheManager(ABC):
181
182
Args:
183
request_id: The request ID.
184
- block_hashes: The block hashes of the request.
184
+ num_running_requests: The number of requests in the RUNNING state.
185
186
Returns:
187
The number of common prefix blocks.