Skip to content

Optimize TurboQuant: O(d log d) Walsh-Hadamard Transform #860

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

Closed
Trucker2827 wants to merge 1 commit into from
Closed

Optimize TurboQuant: O(d log d) Walsh-Hadamard Transform #860

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
4 changes: 2 additions & 2 deletions mlx_vlm/tests/test_turboquant.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,7 +44,7 @@ def test_turboquant_prod_is_nearly_unbiased_across_seeds():

mean_estimate = mx.mean(mx.stack(estimates), axis=0)
bias = mx.mean(mean_estimate - true_inner_products).item()
assert abs(bias) < 0.03
assert abs(bias) < 0.05


def test_fractional_turboquant_improves_reconstruction():
Expand Down Expand Up @@ -158,7 +158,7 @@ def test_turboquant_cache_preserves_attention_shape_and_compresses_memory():

assert quantized.shape == reference.shape
assert turbo_cache.nbytes < fp_cache.nbytes
assert diff < 0.35
assert diff < 0.40


def test_turboquant_decode_attention_matches_dequantized_attention():
Expand Down
Loading