← Search

Kyuyeun Kim

2 accepted papers

2025

MimiQ: Low-Bit Data-Free Quantization of Vision Transformers with Encouraging Inter-Head Attention Similarity

AAAI 2025technical

Data-free quantization (DFQ) is a technique that creates a lightweight network from its full-precision counterpart without the original training data, often through a synthetic dataset. Although several DFQ methods have been proposed for vision transformer (ViT) architectures, they fail to achieve e…

2024

Prefixing Attention Sinks can Mitigate Activation Outliers for Large Language Model Quantization

EMNLP 2024main

Despite recent advances in LLM quantization, activation quantization remains to be challenging due to the activation outliers. Conventional remedies, e.g., mixing precisions for different channels, introduce extra overhead and reduce the speedup. In this work, we develop a simple yet effective strat…

Cited by 7SourcePDFScholar