8 papers
LowRankArena: A Standardized Evaluation Platform for SVD-Based LLM Compression
Zishan Shao, Lixun Zhang, Kangning Cui +10
SVD-based low-rank compression has become a fast-growing direction for reducing the memory and computational cost of large language models (LLMs). However, meaningful comparison ac…
DecodeShare: Tracing the Shared Subspace of LLM Decode-Time Decisions
Zishan Shao, Lixun Zhang, Kangning Cui +10
Large language models (LLMs) handle many tasks with one set of parameters, but under KV-cached inference it is unclear what task-general structure, if any, is used at decode time r…
CLARITree: Cholesky and Lookahead Accelerations for Regression with Interpretable Piecewise Linear Trees
Yixiao Wang, Hayden McTavish, Varun Babbar +2
Regression trees are among the most interpretable yet expressive model classes in machine learning. Historically, greedy induction has been the dominant approach for constructing w…
FlashSVD v1.5: Making Low-Rank Transformers Inference Actually Fast
Wenhao Wu, Zishan Shao, Kangning Cui +5
SVD-based Low-rank compression reduces transformer parameters and nominal FLOPs, but these savings often translate poorly into real LLM serving speedups. We show that this gap is l…
ZEUS: Accelerating Diffusion Models with Only Second-Order Predictor
Yixiao Wang, Ting Jiang, Zishan Shao +6
Denoising generative models deliver high-fidelity generation but remain bottlenecked by inference latency due to the many iterative denoiser calls required during sampling. Trainin…
Enhanced Cyclic Coordinate Descent Methods for Elastic Net Penalized Linear Models
Yixiao Wang, Zishan Shao, Ting Jiang +1
We present a novel enhanced cyclic coordinate descent (ECCD) framework for solving generalized linear models with elastic net constraints that reduces training time in comparison t…
FlashSVD: Memory-Efficient Inference with Streaming for Low-Rank Models
Zishan Shao, Yixiao Wang, Qinsi Wang +6
Singular Value Decomposition (SVD) has recently seen a surge of interest as a simple yet powerful tool for large language models (LLMs) compression, with a growing number of works…
SADA: Stability-guided Adaptive Diffusion Acceleration
Ting Jiang, Yixiao Wang, Hancheng Ye +7
Diffusion models have achieved remarkable success in generative tasks but suffer from high computational costs due to their iterative sampling process and quadratic attention costs…