Improving Neural Network Quantization without Retraining using Outlier Channel Splitting
This work addresses the problem of efficient neural network deployment on commodity hardware for practitioners by providing a no-retraining quantization method, though it is incremental as it builds on prior outlier-handling techniques.
The paper tackles the problem of quantizing floating-point neural networks without retraining by addressing outlier distributions that challenge linear quantization grids, proposing outlier channel splitting (OCS) which duplicates and halves outlier channels to move them toward the distribution center, and shows it outperforms state-of-the-art clipping techniques on ImageNet classification and language modeling with minor overhead.
Quantization can improve the execution latency and energy efficiency of neural networks on both commodity GPUs and specialized accelerators. The majority of existing literature focuses on training quantized DNNs, while this work examines the less-studied topic of quantizing a floating-point model without (re)training. DNN weights and activations follow a bell-shaped distribution post-training, while practical hardware uses a linear quantization grid. This leads to challenges in dealing with outliers in the distribution. Prior work has addressed this by clipping the outliers or using specialized hardware. In this work, we propose outlier channel splitting (OCS), which duplicates channels containing outliers, then halves the channel values. The network remains functionally identical, but affected outliers are moved toward the center of the distribution. OCS requires no additional training and works on commodity hardware. Experimental evaluation on ImageNet classification and language modeling shows that OCS can outperform state-of-the-art clipping techniques with only minor overhead.