2 papers
cs.LG2026
mHC-lite: You Don't Need 20 Sinkhorn-Knopp Iterations
Yongyi Yang, Jianyang Gao
Hyper-Connections (HC) generalizes residual connections by introducing dynamic residual matrices that mix information across multiple residual streams, accelerating convergence in…
cs.LG2025
RaanA: A Fast, Flexible, and Data-Efficient Post-Training Quantization Algorithm
Yongyi Yang, Jianyang Gao, Wei Hu
Post-training Quantization (PTQ) has become a widely used technique for improving inference efficiency of large language models (LLMs). However, existing PTQ methods generally suff…