Distilling Diversity and Control in Diffusion Models
April 28, 2025 ยท View on GitHub
Project Website | ArXiv Preprint
Official code implementation for "Distilling Diversity and Control in Diffusion Models".
Overview
Distilled diffusion models generate images in far fewer timesteps but suffer from "mode collapse" - producing similar outputs despite different random seeds. Our work addresses this critical limitation through:
-
Control Distillation: We discover that control mechanisms (Concept Sliders, LoRAs, DreamBooth) trained on base models can be directly applied to distilled models without retraining.
-
DT-Visualization: A novel analysis technique that reveals what diffusion models "think" the final image will be at intermediate steps.
-
Diversity Distillation: A hybrid inference approach using the base model for only the first timestep before switching to the distilled model, restoring diversity while maintaining speed.
Setup
conda create -n distillation python=3.9
conda activate distillation
git clone https://github.com/rohitgandikota/distillation.git
cd distillation
pip install -r requirements.txt
DT-Visualization
For DT-Visualization - see dt-visualization.ipynb notebook
Diversity Distillation
To improve the diversity of your distilled model - use the notebook diversity_distillation.ipynb
For generating multiple images using a bash script, use the following python evaluation script
python evalscripts/diversity_distillation_sdxl.py --distillation_type 'dmd' --prompts_path 'data/coco_30k.csv' --exp_name 'dmd_diversity_distillation' --device 'cuda:0'
Citing our work
@article{gandikota2025distilling,
title={Distilling Diversity and Control in Diffusion Models},
author={Rohit Gandikota and David Bau},
journal={arXiv preprint arXiv:2503.10637}
year={2025}
}