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 16, 00:45 UTC
Properly check if all fused layers are in the list of targets (#12666)
Thanks @kylesayrs for catching this!
Eldar Kurtic committed
Feb 3, 2025 at 06:42 UTC
c5932e5daceba8a5eefece98f1a769ddf84a864b
1 file changed
+1
-1
vllm/model_executor/layers/quantization/compressed_tensors/utils.py
+1
-1
@@ -236,7 +236,7 @@ def _match_fused_layer(layer_name: str,
236
for type_suffix in possible_layer_types)
237
238
if is_same_parent and is_matching_type and all(
239
- '.'.join([parent_path, type_suffix])
239
+ (f"{parent_path}.{type_suffix}" in target_layers)
240
for type_suffix in possible_layer_types):
241
return target
242