# Hacking Generative Models with Differentiable Network Bending

arXiv: https://arxiv.org/abs/2310.04816 · HTML: https://mlsanity.com/paper/2310.04816/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 6 of its 11 references that are in the map.
- First posted on arXiv 7 Oct 2023.

## Builds on (most cited first, 6 of 6)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis](https://mlsanity.com/paper/2101.04775/) (2021)
- [Monotonic Differentiable Sorting Networks](https://mlsanity.com/paper/2203.09630/) (2022)
- [Towards Counterfactual Image Manipulation via CLIP](https://mlsanity.com/paper/2207.02812/) (2022)

## Built on by (most cited first, 3 of 3)

- [Controlling the image generation process with parametric activation functions](https://mlsanity.com/paper/2510.15778/) (2025)
- [AttentionBender: Manipulating Cross-Attention in Video Diffusion Transformers as a Creative Probe](https://mlsanity.com/paper/2604.20936/) (2026)
- [Unboxing Diffusion Models for the Arts: Interactive Model Bending and Practice-Based Explainability](https://mlsanity.com/paper/2607.22428/) (2026)

## Abstract (arXiv)

In this work, we propose a method to 'hack' generative models, pushing their outputs away from the original training distribution towards a new objective. We inject a small-scale trainable module between the intermediate layers of the model and train it for a low number of iterations, keeping the rest of the network frozen. The resulting output images display an uncanny quality, given by the tension between the original and new objectives that can be exploited for artistic purposes.