xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-07-25 05:57:09 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Fix shared storage connector load kv only load attention layer (#21428)

Browse Source 
Signed-off-by: David Chen <530634352@qq.com>
This commit is contained in:
WeiQing Chen 2025-07-26 22:07:40 +08:00 committed by GitHub
parent a40a8506df
commit 97d6c30cc9
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 10 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
vllm/distributed/kv_transfer/kv_connector/v1/shared_storage_connector.py
View File

@ -156,8 +156,16 @@ class SharedStorageConnector(KVConnectorBase_V1):
logger.info("Inject KV cache of %d tokens to the paged memory", logger.info("Inject KV cache of %d tokens to the paged memory",
len(request.slot_mapping)) len(request.slot_mapping))
for layer_name in forward_context.no_compile_layers: for layer_name in forward_context.no_compile_layers:
attn_layer = forward_context.no_compile_layers[layer_name] layer = forward_context.no_compile_layers[layer_name]
kv_cache_layer = attn_layer.kv_cache[\
# Only process layers that have kv_cache
# attribute (attention layers) Skip non-attention
# layers like FusedMoE/MLP etc.
kv_cache_attr = getattr(layer, 'kv_cache', None)
if kv_cache_attr is None:
continue
kv_cache_layer = kv_cache_attr[ \
forward_context.virtual_engine] forward_context.virtual_engine]
filename = self._generate_filename_debug( filename = self._generate_filename_debug(