Skip to content

[FP8 3/5] Integrate FP8 attention and caching into TabICLv2 - #879

Draft
aw471 wants to merge 11 commits into
fp8-stack/kernelsfrom
fp8-stack/01-tabiclv2
Draft

aw471 wants to merge 11 commits into
fp8-stack/kernelsfrom
fp8-stack/01-tabiclv2

Enable FP8 attention on CUDA architectures with native FP8 support

6c4171b
Select commit
Loading
Failed to load commit list.
Codecov / codecov/project succeeded Nov 16, 2026 in 0s

82.16% (-0.02%) compared to 66b11a3, passed because coverage increased by 0.01% when compared to adjusted base (82.14%)

View this Pull Request on Codecov

82.16% (-0.02%) compared to 66b11a3, passed because coverage increased by 0.01% when compared to adjusted base (82.14%)

Details

Codecov Report

❌ Patch coverage is 85.71429% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.16%. Comparing base (66b11a3) to head (6c4171b).
✅ All tests successful. No failed tests found.

Files with missing lines Patch % Lines
sdm/nn/attention.py 77.77% 6 Missing ⚠️
Additional details and impacted files
@@                  Coverage Diff                  @@
##           fp8-stack/kernels     #879      +/-   ##
=====================================================
- Coverage              82.17%   82.16%   -0.02%     
=====================================================
  Files                    176      177       +1     
  Lines                  11488    11515      +27     
=====================================================
+ Hits                    9440     9461      +21     
- Misses                  2048     2054       +6     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.