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
Add Dockerfile with torch installed, enhance image build logic to check for changes (#1811)
694ad39 chen, suyue committed 6 months ago
Fix mixed-precision accuracy regression when AutoScheme runs with CPU offloading and Hadamard rotation enabled (#1753)
65ae39e Liang Lv committed 6 months ago
Fix GGUF-K RTN routing and fake eval dtype normalization (#1807)
351ba32 Heng Guo committed 6 months ago
[MTP]split gate_up_proj and fix accu gap in rtn quantization (#1758)
c859fab Xin He committed 6 months ago