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
Delay materializing the replaced model weights until quantization (#1307)
d9d20a1 Yi Liu committed 9 months ago
Update publication dates in README and add recipe configuration detai… (#1329)
46224d4 Xin He committed 10 months ago
[GGUF] using quant_nontext_module to control whether quant vision model (#1317)
24e4397 Heng Guo committed 10 months ago
add blockwise way in pure rtn to ease module replacements (#1320)
57d1a8a Wenhua Cheng committed 10 months ago
Add issue templates and PR template with required fields (#1309)
eaf5885 Copilot committed 10 months ago
Enable q_scale calibration for deepseek v2 on Gaudi (#1299)
d8ed35a Wang, Mengni committed 10 months ago
refine moe modellings to release orginal expert module's ram (#1265)
e383eea Weiwei committed 10 months ago
gguf fix bug of tensors not on same tensor and gguf:q2_k_mixed (#1263)
2c67712 Heng Guo committed 10 months ago
GGUF format add support for MoE models with non-linear expert layers. (#1244)
c7d6aee Heng Guo committed 10 months ago