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 15, 08:30 UTC
[MISC] correct signature for LoaderFunction (#18670)
Signed-off-by: Andy Xie <andy.xning@gmail.com>
Ning Xie committed
May 25, 2025 at 11:17 UTC
6c6dcd8611933fb2dd1842b26e63c06a9b018e4f
1 file changed
+1
-1
vllm/model_executor/model_loader/weight_utils.py
+1
-1
@@ -634,7 +634,7 @@ def row_parallel_weight_loader(param: torch.Tensor,
634
return default_weight_loader(param, loaded_weight)
635
636
637
-LoaderFunction = Callable[[torch.Tensor, torch.Tensor], torch.Tensor]
637
+LoaderFunction = Callable[[torch.Tensor, torch.Tensor], None]
638
639
640
def sharded_weight_loader(shard_axis: int) -> LoaderFunction: