Skip to content

[OMNIML-4998] Per-expert weight quantization on TEGroupedLinear#1671

Closed
hychiang-git wants to merge 16 commits into
mainfrom
hungyuehc/omniml-4998-umbrella
Closed

[OMNIML-4998] Per-expert weight quantization on TEGroupedLinear#1671
hychiang-git wants to merge 16 commits into
mainfrom
hungyuehc/omniml-4998-umbrella

[OMNIML-5072] Triton fwd: round-half-to-even via libdevice.rint

1080e68
Select commit
Loading
Failed to load commit list.
Codecov / codecov/project failed Jun 12, 2026 in 1s

75.33% (-2.16%) compared to 2c52e7b

View this Pull Request on Codecov

75.33% (-2.16%) compared to 2c52e7b

Details

Codecov Report

❌ Patch coverage is 0% with 203 lines in your changes missing coverage. Please review.
✅ Project coverage is 75.33%. Comparing base (2c52e7b) to head (1080e68).
⚠️ Report is 35 commits behind head on main.

Files with missing lines Patch % Lines
...rnels/quantization/gemm/grouped_axis0_fakequant.py 0.00% 84 Missing ⚠️
...t/torch/quantization/plugins/transformer_engine.py 0.00% 62 Missing ⚠️
modelopt/torch/quantization/plugins/megatron.py 0.00% 48 Missing ⚠️
modelopt/torch/quantization/model_calib.py 0.00% 8 Missing ⚠️
...delopt/torch/kernels/quantization/gemm/__init__.py 0.00% 1 Missing ⚠️
Additional details and impacted files 
@@            Coverage Diff             @@
##             main    #1671      +/-   ##
==========================================
- Coverage   77.48%   75.33%   -2.16%     
==========================================
  Files         489      512      +23     
  Lines       54415    60305    +5890     
==========================================
+ Hits        42165    45432    +3267     
- Misses      12250    14873    +2623

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

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
View more details on Codecov