Skip to content

add support for per-head attention quantization #442

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

add support for per-head attention quantization #442

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
10 changes: 10 additions & 0 deletions src/compressed_tensors/quantization/lifecycle/initialize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,6 +167,16 @@ def _initialize_scale_zero_point(
# 2. Infer expected scale/zero point shape
if quantization_args.strategy == QuantizationStrategy.TOKEN:
expected_shape = (1, 1)
elif quantization_args.strategy == QuantizationStrategy.ATTN_HEAD:
# supports only GQA models, support others when/if needed
if base_name == "q":
expected_shape = module.config.num_attention_heads
elif base_name in ("k", "v"):
expected_shape = module.config.num_key_value_heads
else:
raise ValueError(
f"Unsupported target {type(module)} for per-attention-head quantization"
)
else:
expected_shape = 1

Expand Down
1 change: 1 addition & 0 deletions src/compressed_tensors/quantization/quant_args.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,6 +101,7 @@ class QuantizationStrategy(str, Enum):
BLOCK = "block"
TOKEN = "token"
TENSOR_GROUP = "tensor_group"
ATTN_HEAD = "attn_head"


class DynamicType(str, Enum):
Expand Down