PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
Commits on main
Fix nvfp4&fp8 packing ram issue, refine all exporting format ram release for origin layer (#1129)
7771268 Yi Liu committed 11 months ago
fix bug of outside block layers do not use best_params (#1128)
ba1c76c Heng Guo committed 11 months ago
fix bug of triton pow wrong data_type when enable torch compile (#1120)
3a5010d Heng Guo committed 11 months ago
Add Python 3.14 support to compatibility test pipeline (#1096)
99bf772 Sun, Xuehao committed 11 months ago
Refactor input normalization by replaying inputs for consistent preprocessing (#1094)
152f1a0 Yi Liu committed 11 months ago