Soft Quantization: Model Compression Via Weight Coupling
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917230924529664 |
|---|---|
| author | Bernstein, Daniel T. Di Carlo, Luca Schwab, David |
| author_facet | Bernstein, Daniel T. Di Carlo, Luca Schwab, David |
| contents | We show that introducing short-range attractive couplings between the weights of a neural network during training provides a novel avenue for model quantization. These couplings rapidly induce the discretization of a model's weight distribution, and they do so in a mixed-precision manner despite only relying on two additional hyperparameters. We demonstrate that, within an appropriate range of hyperparameters, our "soft quantization'' scheme outperforms histogram-equalized post-training quantization on ResNet-20/CIFAR-10. Soft quantization provides both a new pipeline for the flexible compression of machine learning models and a new tool for investigating the trade-off between compression and generalization in high-dimensional loss landscapes. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_21219 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Soft Quantization: Model Compression Via Weight Coupling Bernstein, Daniel T. Di Carlo, Luca Schwab, David Machine Learning Disordered Systems and Neural Networks We show that introducing short-range attractive couplings between the weights of a neural network during training provides a novel avenue for model quantization. These couplings rapidly induce the discretization of a model's weight distribution, and they do so in a mixed-precision manner despite only relying on two additional hyperparameters. We demonstrate that, within an appropriate range of hyperparameters, our "soft quantization'' scheme outperforms histogram-equalized post-training quantization on ResNet-20/CIFAR-10. Soft quantization provides both a new pipeline for the flexible compression of machine learning models and a new tool for investigating the trade-off between compression and generalization in high-dimensional loss landscapes. |
| title | Soft Quantization: Model Compression Via Weight Coupling |
| topic | Machine Learning Disordered Systems and Neural Networks |
| url | https://arxiv.org/abs/2601.21219 |