← Search

Zengke Liu

2 accepted papers

2025

FPE2M2: Approaching Lossless and Efficient Quantization with Native Floating Point

ACL 2025finding

Auto-regressive decoding is a memory-bound job, meaning decoding inference performance is limited by the bandwidth rather than the computational capabilities of the GPU. Weight-only quantization is a promising method to address the memory-bound limitations. Previous studies have followed one of two…

Cited by 0SourcePDFScholar
2025

Rotated Runtime Smooth: Training-Free Activation Smoother for accurate INT4 inference

ICLR 2025poster

Large language models have demonstrated promising capabilities upon scaling up parameters. However, serving large language models incurs substantial computation and memory movement costs due to their large scale. Quantization methods have been employed to reduce service costs and latency. Neverthele…

Cited by 0SourcePDFScholar