From 9480ae24e38cb73c5b665f5843ebd92c75a2039f Mon Sep 17 00:00:00 2001 From: Kyuyeun Kim <62023335+kyuyeunk@users.noreply.github.com> Date: Tue, 2 Sep 2025 10:56:31 -0700 Subject: [PATCH] [Bugfix] Fix packed_factor missing attribute error (#23902) Signed-off-by: Kyuyeun Kim --- vllm/model_executor/layers/linear.py | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/vllm/model_executor/layers/linear.py b/vllm/model_executor/layers/linear.py index 19ff63145024f..f24c87dbf4509 100644 --- a/vllm/model_executor/layers/linear.py +++ b/vllm/model_executor/layers/linear.py @@ -722,8 +722,8 @@ class MergedColumnParallelLinear(ColumnParallelLinear): # If quantized, we need to adjust the offset and size to account # for the packing. if packed_dim == output_dim: - shard_size = shard_size // param.pack_factor - shard_offset = shard_offset // param.pack_factor + shard_size = shard_size // param.packed_factor + shard_offset = shard_offset // param.packed_factor # Special case for Marlin. shard_size, shard_offset = adjust_marlin_shard( param, shard_size, shard_offset) @@ -756,8 +756,8 @@ class MergedColumnParallelLinear(ColumnParallelLinear): # for the packing. packed_dim = getattr(param, "packed_dim", None) if packed_dim == output_dim: - shard_size = shard_size // param.pack_factor - shard_offset = shard_offset // param.pack_factor + shard_size = shard_size // param.packed_factor + shard_offset = shard_offset // param.packed_factor # Special case for Marlin. shard_size, shard_offset = adjust_marlin_shard( param, shard_size, shard_offset) @@ -1107,8 +1107,8 @@ class QKVParallelLinear(ColumnParallelLinear): # If quantized, we need to adjust the offset and size to account # for the packing. if packed_dim == output_dim: - shard_size = shard_size // param.pack_factor - shard_offset = shard_offset // param.pack_factor + shard_size = shard_size // param.packed_factor + shard_offset = shard_offset // param.packed_factor # Special case for Marlin. shard_size, shard_offset = adjust_marlin_shard( @@ -1155,8 +1155,8 @@ class QKVParallelLinear(ColumnParallelLinear): # for the packing. packed_dim = getattr(param, "packed_dim", None) if packed_dim == output_dim: - shard_size = shard_size // param.pack_factor - shard_offset = shard_offset // param.pack_factor + shard_size = shard_size // param.packed_factor + shard_offset = shard_offset // param.packed_factor # Special case for Marlin. shard_size, shard_offset = adjust_marlin_shard(