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
handling transformers version compatibility in lmhead export, bugfix (#130)
4db22e1 WeiweiZhang1 committed 2 years ago
fix falcon quant issue with disable_trust_remote_code (#126)
e2985fd WeiweiZhang1 committed 2 years ago
support real lm-head quantization and mixed precision inference (#114)
4d1caeb wenhuach21 committed 2 years ago
Adjust the default evaluation data type by selecting from the model path configuration (#107)
c42eaa9 WeiweiZhang1 committed 2 years ago
1.8X speedup by disable_low_gpu_mem_usage and reduce memory usage by avoid using torch.cat (#106)
23b60c3 wenhuach21 committed 2 years ago