# DNQ: Dynamic Network Quantization

arXiv: https://arxiv.org/abs/1812.02375 · HTML: https://mlsanity.com/paper/1812.02375/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 6 Dec 2018.

## Built on by (most cited first, 7 of 7)

- [CPT: Efficient Deep Neural Network Training via Cyclic Precision](https://mlsanity.com/paper/2101.09868/) (2021)
- [An Efficient and Scalable Collection of Fly-inspired Voting Units for Visual Place Recognition in Changing Environments](https://mlsanity.com/paper/2109.10986/) (2021)
- [Compacting Deep Neural Networks for Internet of Things: Methods and Applications](https://mlsanity.com/paper/2103.11083/) (2021)
- [DQT: Dynamic Quantization Training via Dequantization-Free Nested Integer Arithmetic](https://mlsanity.com/paper/2508.09176/) (2025)
- [Better Schedules for Low Precision Training of Deep Neural Networks](https://mlsanity.com/paper/2403.02243/) (2024)
- [TTQ: Activation-Aware Test-Time Quantization to Accelerate LLM Inference On The Fly](https://mlsanity.com/paper/2603.19296/) (2026)
- [Assessing the Potential for Catastrophic Failure in Dynamic Post-Training Quantization](https://mlsanity.com/paper/2510.02457/) (2025)

## Abstract (arXiv)

Network quantization is an effective method for the deployment of neural networks on memory and energy constrained mobile devices. In this paper, we propose a Dynamic Network Quantization (DNQ) framework which is composed of two modules: a bit-width controller and a quantizer. Unlike most existing quantization methods that use a universal quantization bit-width for the whole network, we utilize policy gradient to train an agent to learn the bit-width of each layer by the bit-width controller. This controller can make a trade-off between accuracy and compression ratio. Given the quantization bit-width sequence, the quantizer adopts the quantization distance as the criterion of the weights importance during quantization. We extensively validate the proposed approach on various main-stream neural networks and obtain impressive results.