Identifying Sensitive Weights via Post-quantization Integral
By: Yuezhou Hu , Weiyu Huang , Zichen Liang and more
Potential Business Impact:
Makes big computer brains run faster and cheaper.
Serving Large Language Models (LLMs) is costly. However, post-training weight quantization can address this problem by both compressing their sizes for limited memory and saving bandwidth for acceleration. As not all weight dimensions are equally important, those methods typically rely on a sensitivity metric, which indicates the element-wise influence of weights on loss function and is used to preprocess original weights for better quantization. In this work, we conduct an empirical study on the accuracy of the sensitivity metric, and find that existing gradient and Hessian based metrics are very inaccurate: they underestimate quantization's impact on the loss function by orders of magnitude, mainly due to the small convergence radius of local 2nd order approximation, \ie, gradient and Hessian term in Taylor's formula. To tackle this problem, we propose Post-quantization Integral (PQI), an accurate metric to estimate posterior sensitivity in a fine-grained manner. To leverage this accurate metric, we further propose ReQuant, a simple yet powerful framework that mainly consists of two Dense-and-Sparse detach components: self-adaptive outlier selection and step-wise significant weights detach. Results show that ReQuant boosts state-of-the-art post-training quantization methods, with a pronounced improvement of 2.66 perplexity gain on Llama 3.2 1B with QTIP.
Similar Papers
Sensitivity-Aware Post-Training Quantization for Deep Neural Networks
CV and Pattern Recognition
Makes smart computer programs smaller, faster, and still accurate.
Towards Superior Quantization Accuracy: A Layer-sensitive Approach
Machine Learning (CS)
Makes smart computer brains smaller and faster.
Exploring Layer-wise Information Effectiveness for Post-Training Quantization in Small Language Models
Machine Learning (CS)
Makes smart computer programs smaller and faster.