QGen: On the Ability to Generalize in Quantization Aware Training
This work addresses the generalization issue in quantized models for machine learning practitioners, providing theoretical insights and empirical validation, though it is incremental as it builds on existing connections between sharpness and generalization.
The paper tackles the generalization properties of quantized neural networks, demonstrating that quantization acts as regularization and deriving a bound for generalization based on quantization noise, validated by experiments on over 2000 models across CIFAR-10, CIFAR-100, and ImageNet datasets.
Quantization lowers memory usage, computational requirements, and latency by utilizing fewer bits to represent model weights and activations. In this work, we investigate the generalization properties of quantized neural networks, a characteristic that has received little attention despite its implications on model performance. In particular, first, we develop a theoretical model for quantization in neural networks and demonstrate how quantization functions as a form of regularization. Second, motivated by recent work connecting the sharpness of the loss landscape and generalization, we derive an approximate bound for the generalization of quantized models conditioned on the amount of quantization noise. We then validate our hypothesis by experimenting with over 2000 models trained on CIFAR-10, CIFAR-100, and ImageNet datasets on convolutional and transformer-based models.