# And the Bit Goes Down: Revisiting the Quantization of Neural Networks Source: [And the Bit Goes Down: Revisiting the Quantization of Neural Networks]() Collection index: [llms.txt]() Data updated: 2025-10-25T22:43:58.904Z Authors: Pierre Stock, Armand Joulin, R\\'emi Gribonval, Benjamin Graham, Herv\\'e J\\'egou Published: 2019-07-12T11:52:54.000Z ## Abstract In this paper, we address the problem of reducing the memory footprint of convolutional network architectures. We introduce a vector quantization method that aims at preserving the quality of the reconstruction of the network outputs rather than its weights. The principle of our approach is that it minimizes the loss reconstruction error for in-domain inputs. Our method only requires a set of unlabelled data at quantization time and allows for efficient inference on CPU by using byte-aligned codebooks to store the compressed weights. We validate our approach by quantizing a high performing ResNet-50 model to a memory size of 5MB (20x compression factor) while preserving a top-1 accuracy of 76.1% on ImageNet object classification and by compressing a Mask R-CNN with a 26x factor. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Image Classification on ImageNet (val) (CR and Accuracy Loss)](): Image Classification; ImageNet (val); Contrastive Ratio (CR): 26. ## Original sources and implementation [Original paper]() [PDF]()