# Rethinking Token-wise Feature Caching: Accelerating Diffusion Transformers with Dual Feature Caching

arXiv: https://arxiv.org/abs/2412.18911 · HTML: https://mlsanity.com/paper/2412.18911/

## Key facts

- Cited by 28 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- Builds on 25 of its 44 references that are in the map.
- First posted on arXiv 25 Dec 2024.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)

## Builds on (most cited first, 25 of 25)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://mlsanity.com/paper/2205.11487/) (2022)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow](https://mlsanity.com/paper/2209.03003/) (2022)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets](https://mlsanity.com/paper/2311.15127/) (2023)
- [DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps](https://mlsanity.com/paper/2206.00927/) (2022)
- [VBench: Comprehensive Benchmark Suite for Video Generative Models](https://mlsanity.com/paper/2311.17982/) (2023)
- [ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generation](https://mlsanity.com/paper/2304.05977/) (2023)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models](https://mlsanity.com/paper/2211.01095/) (2022)
- [PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis](https://mlsanity.com/paper/2310.00426/) (2023)
- [DeepCache: Accelerating Diffusion Models for Free](https://mlsanity.com/paper/2312.00858/) (2023)
- [Token Merging for Fast Stable Diffusion](https://mlsanity.com/paper/2303.17604/) (2023)
- [Delta-DiT: A Training-Free Acceleration Method Tailored for Diffusion Transformers](https://mlsanity.com/paper/2406.01125/) (2024)
- [SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models](https://mlsanity.com/paper/2411.05007/) (2024)
- [SageAttention: Accurate 8-Bit Attention for Plug-and-play Inference Acceleration](https://mlsanity.com/paper/2410.02367/) (2024)
- [Accelerating Diffusion Transformers with Token-wise Feature Caching](https://mlsanity.com/paper/2410.05317/) (2024)
- [Real-Time Video Generation with Pyramid Attention Broadcast](https://mlsanity.com/paper/2408.12588/) (2024)
- [Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching](https://mlsanity.com/paper/2406.01733/) (2024)
- [FORA: Fast-Forward Caching in Diffusion Transformer Acceleration](https://mlsanity.com/paper/2407.01425/) (2024)
- [SageAttention2: Efficient Attention with Thorough Outlier Smoothing and Per-thread INT4 Quantization](https://mlsanity.com/paper/2411.10958/) (2024)
- [DiT4Edit: Diffusion Transformer for Image Editing](https://mlsanity.com/paper/2411.03286/) (2024)

## Built on by (most cited first, 28 of 28)

- [MagCache: Fast Video Generation with Magnitude-Aware Cache](https://mlsanity.com/paper/2506.09045/) (2025)
- [FreqCa: Accelerating Diffusion Models via Frequency-Aware Caching](https://mlsanity.com/paper/2510.08669/) (2025)
- [Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching](https://mlsanity.com/paper/2509.10312/) (2025)
- [ERTACache: Error Rectification and Timesteps Adjustment for Efficient Diffusion](https://mlsanity.com/paper/2508.21091/) (2025)
- [Astraea: A Token-wise Acceleration Framework for Video Diffusion Transformers](https://mlsanity.com/paper/2506.05096/) (2025)
- [TaoCache: Structure-Maintained Video Generation Acceleration](https://mlsanity.com/paper/2508.08978/) (2025)
- [Let Features Decide Their Own Solvers: Hybrid Feature Caching for Diffusion Transformers](https://mlsanity.com/paper/2510.04188/) (2025)
- [Chipmunk: Training-Free Acceleration of Diffusion Transformers with Dynamic Column-Sparse Deltas](https://mlsanity.com/paper/2506.03275/) (2025)
- [Efficient Video Diffusion Models: Advancements and Challenges](https://mlsanity.com/paper/2604.15911/) (2026)
- [Forecasting When to Forecast: Accelerating Diffusion Models with Confidence-Gated Taylor](https://mlsanity.com/paper/2508.02240/) (2025)
- [RegionE: Adaptive Region-Aware Generation for Efficient Image Editing](https://mlsanity.com/paper/2510.25590/) (2025)
- [FlashOmni: A Unified Sparse Attention Engine for Diffusion Transformers](https://mlsanity.com/paper/2509.25401/) (2025)
- [No Cache Left Idle: Accelerating diffusion model via Extreme-slimming Caching](https://mlsanity.com/paper/2512.12604/) (2025)
- [Beyond Fixed Formulas: Data-Driven Linear Predictor for Efficient Diffusion Models](https://mlsanity.com/paper/2604.26365/) (2026)
- [WorldDynCache: Risk-Controlled Latent Dynamics Approximation for Diffusion World Model](https://mlsanity.com/paper/2608.01845/) (2026)
- [LoSA: Near-Lossless Sparse Attention for Training-Free Video Diffusion Acceleration](https://mlsanity.com/paper/2608.12032/) (2026)
- [InvarDiff: Cross-Scale Invariance Caching for Accelerated Diffusion Models](https://mlsanity.com/paper/2512.05134/) (2025)
- [NanoFLUX: Distillation-Driven Compression of Large Text-to-Image Generation Models for Mobile Devices](https://mlsanity.com/paper/2602.06879/) (2026)
- [Dual-Rate Diffusion: Accelerating diffusion models with an interleaved heavy-light network](https://mlsanity.com/paper/2605.18190/) (2026)
- [MoECa: Aligning Feature Reuse with Expert Decomposition in Diffusion Transformers](https://mlsanity.com/paper/2606.15615/) (2026)
- [Token-Operations-Oriented Inference Optimization Techniques for Large Models](https://mlsanity.com/paper/2606.20295/) (2026)
- [DriveCache: Action-Aware Caching for Driving World Model Inference](https://mlsanity.com/paper/2608.16354/) (2026)
- [AViTS: Adaptive Spatiotemporal Token Selection for Efficient Dynamic-Resolution Generation](https://mlsanity.com/paper/2608.17995/) (2026)
- [Memory-Efficient Training-Free Acceleration of Diffusion Transformers with BaryCache](https://mlsanity.com/paper/2608.28670/) (2026)
- [GeoShrink: Accelerating Diffusion Transformers with Two Lines of Code](https://mlsanity.com/paper/2609.33723/) (2026)
- [CAST: Reconstruction-Coupled Acceleration of Interactive World Models](https://mlsanity.com/paper/2609.34144/) (2026)
- [DeCoPrune: Efficient KV-Cache Pruning for Autoregressive Video Diffusion via Denoising Consistency](https://mlsanity.com/paper/2609.39096/) (2026)
- [Frequency-Step Cache: Separating Synchronization and Repair for Efficient Autoregressive Video Diffusion](https://mlsanity.com/paper/2610.15922/) (2026)

## Abstract (arXiv)

Diffusion Transformers (DiT) have become the dominant methods in image and video generation yet still suffer substantial computational costs. As an effective approach for DiT acceleration, feature caching methods are designed to cache the features of DiT in previous timesteps and reuse them in the next timesteps, allowing us to skip the computation in the next timesteps. Among them, token-wise feature caching has been introduced to perform different caching ratios for different tokens in DiTs, aiming to skip the computation for unimportant tokens while still computing the important ones. In this paper, we propose to carefully check the effectiveness in token-wise feature caching with the following two questions: (1) Is it really necessary to compute the so-called "important" tokens in each step? (2) Are so-called important tokens really important? Surprisingly, this paper gives some counter-intuition answers, demonstrating that consistently computing the selected ``important tokens'' in all steps is not necessary. The selection of the so-called ``important tokens'' is often ineffective, and even sometimes shows inferior performance than random selection. Based on these observations, this paper introduces dual feature caching referred to as DuCa, which performs aggressive caching strategy and conservative caching strategy iteratively and selects the tokens for computing randomly. Extensive experimental results demonstrate the effectiveness of our method in DiT, PixArt, FLUX, and OpenSora, demonstrating significant improvements than the previous token-wise feature caching.