Skip to content

feat(quant): add constant_amax to pin activation input_scale (NVFP4 experts input_scale=1.0)#1947

Merged
Edwardf0t1 merged 2 commits into
mainfrom
feat/nvfp4-constant-amax-input-scale
Jul 10, 2026
Merged

feat(quant): add constant_amax to pin activation input_scale (NVFP4 experts input_scale=1.0)#1947
Edwardf0t1 merged 2 commits into
mainfrom
feat/nvfp4-constant-amax-input-scale

fix(quant): align constant_amax buffer device + forbid dual amax modes

4e293de
Select commit
Loading
Failed to load commit list.
Codecov / codecov/project failed Jul 10, 2026 in 1s

74.76% (-3.04%) compared to 0593df0

View this Pull Request on Codecov

74.76% (-3.04%) compared to 0593df0

Details

Codecov Report

❌ Patch coverage is 94.73684% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 74.76%. Comparing base (0593df0) to head (4e293de).
⚠️ Report is 2 commits behind head on main.

Files with missing lines Patch % Lines
.../torch/quantization/nn/modules/tensor_quantizer.py 85.71% 1 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main    #1947      +/-   ##
==========================================
- Coverage   77.80%   74.76%   -3.04%     
==========================================
  Files         519      519              
  Lines       57960    57977      +17     
==========================================
- Hits        45095    43348    -1747     
- Misses      12865    14629    +1764     

☔ 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.