Research map / Architectures and efficiency / RNNs and normalization
Diffusion on Demand: Selective Caching and Modulation for Efficient Generation
Hee Min Choi, Hyoa Kang, Dokwan Oh, Nam Ik Cho
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes a selective caching and modulation approach for diffusion transformers to reduce inference cost beyond binary cache-or-compute feature caching.
Problem
Diffusion transformers are computationally expensive at inference; existing feature caching methods only choose between reusing cached features and fully recomputing the model, limiting efficiency gains.
Approach
Goes beyond the binary choice of reusing cached features or running the full model by selectively caching features and modulating them, enabling finer-grained computation reuse across timesteps in diffusion transformers.
Experimental setup
Baselines: feature caching methods
Builds on: feature caching
Closest papers by meaning
- Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching (NeurIPS 2024 · similarity 0.92)
- ProCache: Constraint-Aware Feature Caching with Selective Computation for Diffusion Transformer Acceleration (AAAI 2026 · similarity 0.92)
- Accelerating Diffusion Transformers with Token-wise Feature Caching (ICLR 2025 · similarity 0.91)
- Evolutionary Caching to Accelerate Your Off-the-Shelf Diffusion Model (ICLR 2026 · similarity 0.91)
- HarmoniCa: Harmonizing Training and Inference for Better Feature Caching in Diffusion Transformer Acceleration (ICML 2025 · similarity 0.90)
- Cache Me if You Can: Accelerating Diffusion Models through Block Caching (CVPR 2024 · similarity 0.89)
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.
