3 papers
cs.LG2026
When Good Enough Is Optimal: Multiplication-Only Matrix Inversion Approximation for Quantized Gated DeltaNet
Luoming Zhang, Yuwei Ren, Kui Zhang +7
Matrix inversion in chunk-wise parallel linear attention is a major bottleneck for long-context modeling, particularly on NPUs, where forward-substitution-based methods exhibit lim…
cs.LG2024
ZipCache: Accurate and Efficient KV Cache Quantization with Salient Token Identification
Yefei He, Luoming Zhang, Weijia Wu +3
KV cache stores key and value states from previous tokens to avoid re-computation, yet it demands substantial storage space, especially for long sequences. Adaptive KV cache compre…
cs.CV2024
Towards Accurate Post-training Quantization for Reparameterized Models
Luoming Zhang, Yefei He, Wen Fei +4
Model reparameterization is a widely accepted technique for improving inference speed without compromising performance. However, current Post-training Quantization (PTQ) methods of…