Skip to content

GitLab

Unverified Commit 9480ae24 authored by Kyuyeun Kim's avatar Kyuyeun Kim Committed by GitHub
Browse files

[Bugfix] Fix packed_factor missing attribute error (#23902) 


Signed-off-by: default avatarKyuyeun Kim <kyuyeunk@google.com>
parent f399182e
Show whitespace changes
Inline Side-by-side
Showing with 8 additions and 8 deletions
vllm/model_executor/layers/linear.py
View file @ 9480ae24
...@@ -722,8 +722,8 @@ class MergedColumnParallelLinear(ColumnParallelLinear): ...@@ -722,8 +722,8 @@ class MergedColumnParallelLinear(ColumnParallelLinear):
# If quantized, we need to adjust the offset and size to account # If quantized, we need to adjust the offset and size to account
# for the packing. # for the packing.
if packed_dim == output_dim: if packed_dim == output_dim:
shard_size = shard_size // param.pack_factor shard_size = shard_size // param.packed_factor
shard_offset = shard_offset // param.pack_factor shard_offset = shard_offset // param.packed_factor
# Special case for Marlin. # Special case for Marlin.
shard_size, shard_offset = adjust_marlin_shard( shard_size, shard_offset = adjust_marlin_shard(
param, shard_size, shard_offset) param, shard_size, shard_offset)
...@@ -756,8 +756,8 @@ class MergedColumnParallelLinear(ColumnParallelLinear): ...@@ -756,8 +756,8 @@ class MergedColumnParallelLinear(ColumnParallelLinear):
# for the packing. # for the packing.
packed_dim = getattr(param, "packed_dim", None) packed_dim = getattr(param, "packed_dim", None)
if packed_dim == output_dim: if packed_dim == output_dim:
shard_size = shard_size // param.pack_factor shard_size = shard_size // param.packed_factor
shard_offset = shard_offset // param.pack_factor shard_offset = shard_offset // param.packed_factor
# Special case for Marlin. # Special case for Marlin.
shard_size, shard_offset = adjust_marlin_shard( shard_size, shard_offset = adjust_marlin_shard(
param, shard_size, shard_offset) param, shard_size, shard_offset)
...@@ -1107,8 +1107,8 @@ class QKVParallelLinear(ColumnParallelLinear): ...@@ -1107,8 +1107,8 @@ class QKVParallelLinear(ColumnParallelLinear):
# If quantized, we need to adjust the offset and size to account # If quantized, we need to adjust the offset and size to account
# for the packing. # for the packing.
if packed_dim == output_dim: if packed_dim == output_dim:
shard_size = shard_size // param.pack_factor shard_size = shard_size // param.packed_factor
shard_offset = shard_offset // param.pack_factor shard_offset = shard_offset // param.packed_factor
# Special case for Marlin. # Special case for Marlin.
shard_size, shard_offset = adjust_marlin_shard( shard_size, shard_offset = adjust_marlin_shard(
...@@ -1155,8 +1155,8 @@ class QKVParallelLinear(ColumnParallelLinear): ...@@ -1155,8 +1155,8 @@ class QKVParallelLinear(ColumnParallelLinear):
# for the packing. # for the packing.
packed_dim = getattr(param, "packed_dim", None) packed_dim = getattr(param, "packed_dim", None)
if packed_dim == output_dim: if packed_dim == output_dim:
shard_size = shard_size // param.pack_factor shard_size = shard_size // param.packed_factor
shard_offset = shard_offset // param.pack_factor shard_offset = shard_offset // param.packed_factor
# Special case for Marlin. # Special case for Marlin.
shard_size, shard_offset = adjust_marlin_shard( shard_size, shard_offset = adjust_marlin_shard(
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment