Skip to content

switch llama test scripts to new version of Int4WeightOnlyConfig #3493

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Stonepia wants to merge 1 commit into
base: main
Choose a base branch
from
Open

switch llama test scripts to new version of Int4WeightOnlyConfig #3493

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 8 additions & 1 deletion torchao/_models/llama/eval.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,9 +93,16 @@ def run_evaluation(
assert groupsize in [32, 64, 128, 256], (
f"int4wo groupsize needs to be one of [32,64,128,256] but got {groupsize}"
)
int4_packing_format = (
"plain_int32" if device == "xpu" else "tile_packed_to_4d"
)
quantize_(
model.to(device),
Int4WeightOnlyConfig(group_size=groupsize, use_hqq=use_hqq, version=1),
Int4WeightOnlyConfig(
group_size=groupsize,
use_hqq=use_hqq,
int4_packing_format=int4_packing_format,
),
)
if "uintx" in quantization:
# uintx-nbits-groupsize
Expand Down
9 changes: 8 additions & 1 deletion torchao/_models/llama/generate.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -430,9 +430,16 @@ def ffn_or_attn_only(mod, fqn):
], (
f"int4wo group_size needs to be one of [32,64,128,256] but got {group_size}"
)
int4_packing_format = (
"plain_int32" if device == "xpu" else "tile_packed_to_4d"
)
quantize_(
model,
Int4WeightOnlyConfig(group_size=group_size, use_hqq=use_hqq, version=1),
Int4WeightOnlyConfig(
group_size=group_size,
use_hqq=use_hqq,
int4_packing_format=int4_packing_format,
),
)
elif "int4dq-" in quantization:
from torchao.dtypes import CutlassInt4PackedLayout
Expand Down