Skip to content
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

Jump to bottom

Add overflow protection for quantization bias to reduce quantization precision loss #21645

Merged
merged 4 commits into from
Aug 28, 2024
Merged

Add overflow protection for quantization bias to reduce quantization precision loss #21645

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
f810324
Add overflow protection for quantization bias to reduce precision loss
duanshengliu Aug 6, 2024
e3370bc
Merge branch 'microsoft:main' into improve-small-scale-acc
duanshengliu Aug 23, 2024
421d777
Merge branch 'microsoft:main' into improve-small-scale-acc
duanshengliu Aug 23, 2024
015ccae
Merge branch 'microsoft:main' into improve-small-scale-acc
duanshengliu Aug 23, 2024
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
4 changes: 3 additions & 1 deletion onnxruntime/python/tools/quantization/base_quantizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -230,7 +230,9 @@ def quantize_bias_static_impl(self, bias_name, input_scale, weight_scale, beta=1
# TODO: This formula should be explained including why the scale is not estimated for the bias as well.
bias_scale = input_scale * weight_scale * beta

quantized_data = (np.asarray(bias_data) / bias_scale).round().astype(np.int32)
quantized_data = (np.asarray(bias_data) / bias_scale).round()
quantized_data = np.clip(quantized_data, np.iinfo(np.int32).min, np.iinfo(np.int32).max)
quantized_data = quantized_data.astype(np.int32)

# update bias initializer
bias_np_data = np.asarray(quantized_data, dtype=np.int32).reshape(bias_initializer.dims)
Expand Down
Loading