# DynaQuant: Dynamic Mixed-Precision Quantization for Learned Image Compression

arXiv: https://arxiv.org/abs/2511.07903 · HTML: https://mlsanity.com/paper/2511.07903/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 41 references that are in the map.
- First posted on arXiv 11 Nov 2025.

## Builds on (most cited first, 19 of 19)

- [DARTS: Differentiable Architecture Search](https://mlsanity.com/paper/1806.09055/) (2018)
- [Variational image compression with a scale hyperprior](https://mlsanity.com/paper/1802.01436/) (2018)
- [Learned Step Size Quantization](https://mlsanity.com/paper/1902.08153/) (2019)
- [Up or Down? Adaptive Rounding for Post-Training Quantization](https://mlsanity.com/paper/2004.10568/) (2020)
- [Learned Image Compression with Discretized Gaussian Mixture Likelihoods and Attention Modules](https://mlsanity.com/paper/2001.01568/) (2020)
- [HAQ: Hardware-Aware Automated Quantization with Mixed Precision](https://mlsanity.com/paper/1811.08886/) (2018)
- [HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precision](https://mlsanity.com/paper/1905.03696/) (2019)
- [Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning](https://mlsanity.com/paper/2208.11580/) (2022)
- [ELIC: Efficient Learned Image Compression with Unevenly Grouped Space-Channel Contextual Adaptive Coding](https://mlsanity.com/paper/2203.10886/) (2022)
- [HAWQ-V2: Hessian Aware trace-Weighted Quantization of Neural Networks](https://mlsanity.com/paper/1911.03852/) (2019)
- [Learned Image Compression with Mixed Transformer-CNN Architectures](https://mlsanity.com/paper/2303.14978/) (2023)
- [QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training Quantization](https://mlsanity.com/paper/2203.05740/) (2022)
- [Transformer-based Image Compression](https://mlsanity.com/paper/2111.06707/) (2021)
- [C3: High-performance and low-complexity neural compression from a single image or video](https://mlsanity.com/paper/2312.02753/) (2023)
- [MambaVC: Learned Visual Compression with Selective State Spaces](https://mlsanity.com/paper/2405.15413/) (2024)
- [Towards End-to-End Image Compression and Analysis with Transformers](https://mlsanity.com/paper/2112.09300/) (2021)
- [Memory-Efficient Learned Image Compression with Pruned Hyperprior Module](https://mlsanity.com/paper/2206.10083/) (2022)
- [Flexible Mixed Precision Quantization for Learned Image Compression](https://mlsanity.com/paper/2506.01221/) (2025)
- [ShiftLIC: Lightweight Learned Image Compression with Spatial-Channel Shift Operations](https://mlsanity.com/paper/2503.23052/) (2025)

## Built on by (most cited first, 1 of 1)

- [Hadamard-Domain Model Quantization for Learned Image Coding](https://mlsanity.com/paper/2608.01653/) (2026)

## Abstract (arXiv)

Prevailing quantization techniques in Learned Image Compression (LIC) typically employ a static, uniform bit-width across all layers, failing to adapt to the highly diverse data distributions and sensitivity characteristics inherent in LIC models. This leads to a suboptimal trade-off between performance and efficiency. In this paper, we introduce DynaQuant, a novel framework for dynamic mixed-precision quantization that operates on two complementary levels. First, we propose content-aware quantization, where learnable scaling and offset parameters dynamically adapt to the statistical variations of latent features. This fine-grained adaptation is trained end-to-end using a novel Distance-aware Gradient Modulator (DGM), which provides a more informative learning signal than the standard Straight-Through Estimator. Second, we introduce a data-driven, dynamic bit-width selector that learns to assign an optimal bit precision to each layer, dynamically reconfiguring the network's precision profile based on the input data. Our fully dynamic approach offers substantial flexibility in balancing rate-distortion (R-D) performance and computational cost. Experiments demonstrate that DynaQuant achieves rd performance comparable to full-precision models while significantly reducing computational and storage requirements, thereby enabling the practical deployment of advanced LIC on diverse hardware platforms.