xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-17 06:35:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

[DCP] check return_lse for all layers in dcp (#27929)

Browse Source 
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
This commit is contained in:
Chen Zhang 2025-11-05 06:27:25 -08:00 committed by GitHub
parent 0606bea2b6
commit 5d16d0fa62
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/v1/worker/gpu_model_runner.py
View File

@ -4680,10 +4680,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
kv_transfer_group.set_host_xfer_buffer_ops(copy_kv_blocks) kv_transfer_group.set_host_xfer_buffer_ops(copy_kv_blocks)
if self.dcp_world_size > 1: if self.dcp_world_size > 1:
layer_names = self.attn_groups[0][0].layer_names layers = get_layers_from_vllm_config(self.vllm_config, AttentionLayerBase)
layers = get_layers_from_vllm_config(
self.vllm_config, AttentionLayerBase, layer_names
)
for layer in layers.values(): for layer in layers.values():
assert layer.impl.need_to_return_lse_for_decode, ( assert layer.impl.need_to_return_lse_for_decode, (
"DCP requires attention impls to return" "DCP requires attention impls to return"