Research map / Reinforcement learning / RL from feedback
Diffusion Controller: Framework, Algorithms and Parameterization
Tong Yang, Moonkyung Ryu, Chih-Wei Hsu, Guy Tennenholtz, Yuejie Chi, Craig Boutilier, Bo Dai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use LS-MDP-motivated split parameterization instead of naive side network (ablation) | DiffCon (gray-box) · SFT, SD v1.4 | HPS-v2 win rate: 0.5655 -> 0.6667 | Improved |
| use LS-MDP-motivated split parameterization instead of naive side network (ablation) | DiffCon (gray-box) · RWL, SD v1.4 | HPS-v2 win rate: 0.5060 -> 0.6815 | Improved |
| use LS-MDP-motivated split parameterization instead of naive side network (ablation) | DiffCon (gray-box) · PPO, SD v1.4 | HPS-v2 win rate: 0.5201 -> 0.6957 | Improved |
| replace white-box LoRA with gray-box DiffCon (fewer parameters) | DiffCon · SFT, SD v1.4 | HPS-v2 win rate: 0.5766 -> 0.6667 | Improved |
| replace white-box LoRA with gray-box DiffCon | DiffCon · RWL, SD v1.4 | HPS-v2 win rate: 0.6109 -> 0.6815 | Improved |
| replace white-box LoRA with gray-box DiffCon | DiffCon · PPO, SD v1.4 | HPS-v2 win rate: 0.9048 -> 0.6957 | Hurt |
| add DiffCon on top of LoRA (separate training, DiffCon-S) | LoRA · PPO, SD v1.4 | HPS-v2 win rate: 0.9048 -> 0.9315 | Improved |
| add DiffCon on top of LoRA (joint training, DiffCon-J) | LoRA · RWL, SD v1.4 | HPS-v2 win rate: 0.6109 -> 0.6555 | Improved |
| add DiffCon on top of LoRA (separate training, DiffCon-S) | LoRA · RWL, SD v1.4 | HPS-v2 win rate: 0.6109 -> 0.7091 | Improved |
| increase test-time side-network guidance strength (ablation) | DiffCon · SD v1.4 text-to-image | generation quality | Improved |
Closest papers by meaning
- Stochastic Control for Fine-tuning Diffusion Models: Optimality, Regularity, and Convergence (ICML 2025 · similarity 0.92)
- Learning to Sample From Diffusion Models via Inverse Reinforcement Learning (NeurIPS 2026 · similarity 0.91)
- Maximum Entropy Inverse Reinforcement Learning of Diffusion Models with Energy-Based Models (NeurIPS 2024 · similarity 0.90)
- Diffusion Spectral Representation for Reinforcement Learning (NeurIPS 2024 · similarity 0.90)
- Designing Reinforcement Learning for Diffusion Models: A Unified Path-Space View (NeurIPS 2026 · similarity 0.90)
- Adding Conditional Control to Diffusion Models with Reinforcement Learning (ICLR 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
