QuantTune: Optimizing Model Quantization with Adaptive Outlier-Driven Fine Tuning

arXiv:2403.06497v1 Announce Type: new
Abstract: Transformer-based models have gained widespread popularity in both the computer vision (CV) and natural language processing (NLP) fields. However, significant challenges arise during post-training linear quantization, leading to noticeable reductions in inference accuracy. Our study focuses on uncovering the underlying causes of these accuracy drops and proposing a quantization-friendly fine-tuning method, \textbf{QuantTune}. Firstly, our analysis revealed that, on average, 65\% of quantization errors result from the precision loss incurred by the dynamic range amplification effect …

abstract accuracy and natural language processing arxiv challenges computer computer vision cs.cv cs.mm fields however inference language language processing linear natural natural language natural language processing nlp outlier processing quantization study training transformer type vision