Relational Feature Caching for Accelerating Diffusion Transformers

• URL: http://arxiv.org/abs/2602.19506v1
• Date: Mon, 23 Feb 2026 04:45:38 GMT
• Title: Relational Feature Caching for Accelerating Diffusion Transformers
• Abstract summary: Feature caching approaches accelerate diffusion transformers (DiTs) by storing the output features of computationally expensive modules at certain timesteps.<n>Recent forecasting-based caching approaches employ temporal extrapolation techniques to approximate the output features with cached ones.<n>We propose relational feature caching (RFC), a novel framework that leverages the input-output relationship to enhance the accuracy of the feature prediction.
• Score: 19.211788552579655
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Feature caching approaches accelerate diffusion transformers (DiTs) by storing the output features of computationally expensive modules at certain timesteps, and exploiting them for subsequent steps to reduce redundant computations. Recent forecasting-based caching approaches employ temporal extrapolation techniques to approximate the output features with cached ones. Although effective, relying exclusively on temporal extrapolation still suffers from significant prediction errors, leading to performance degradation. Through a detailed analysis, we find that 1) these errors stem from the irregular magnitude of changes in the output features, and 2) an input feature of a module is strongly correlated with the corresponding output. Based on this, we propose relational feature caching (RFC), a novel framework that leverages the input-output relationship to enhance the accuracy of the feature prediction. Specifically, we introduce relational feature estimation (RFE) to estimate the magnitude of changes in the output features from the inputs, enabling more accurate feature predictions. We also present relational cache scheduling (RCS), which estimates the prediction errors using the input features and performs full computations only when the errors are expected to be substantial. Extensive experiments across various DiT models demonstrate that RFC consistently outperforms prior approaches significantly. Project page is available at https://cvlab.yonsei.ac.kr/projects/RFC

Related papers

• Accelerating Diffusion Transformers with Gaussian Process Rectified Feature Cache [17.39547918764814]
  Diffusion Transformers have become the dominant paradigm in generative AI, but their high computational costs hinder real-time applications.<n>Prediction-based feature caching is widely used to accelerate diffusion transformers; however, as the number of steps increases, the deviation between its predictions and the reference full-compute trajectory gradually grows.<n>An intuitive idea is to use an online regression model to dynamically correct this deviation, but it faces the issue of label data being unavailable during the acceleration process.<n>This paper presents a statistical observation that the residuals between the features of full computation steps using caching methods and reference full-compute trajectory locally exhibit a
  arXiv  Detail & Related papers  (2026-09-05T08:54:54Z)
• Prediction of Prediction (PoP): Inter-Layer Activation Fusion for Single-Pass Hallucination Detection in Large Language Models [0.0]
  This work evaluates whether internal hidden-state transition dynamics during generation can signal factual errors without auxiliary decoding calls.<n>We introduce Prediction of Prediction (PoP), a mechanism that captures layer-transition uncertainty.<n>PoP achieves an area under the receiver operating characteristic curve (AUROC) of 75.5% for factual-correctness classification.
  arXiv  Detail & Related papers  (2026-08-27T14:17:14Z)
• Unbiased Gradient Estimation for Event Binning via Functional Backpropagation [64.88399635309918]
  We propose a novel framework for unbiased gradient estimation of arbitrary binning functions by synthesizing weak derivatives during backpropagation.<n>We achieve 9.4% lower EPE in self-supervised optical flow, and 5.1% lower RMS error in SLAM, demonstrating broad benefits for event-based visual perception.
  arXiv  Detail & Related papers  (2026-02-13T04:05:03Z)
• TARFVAE: Efficient One-Step Generative Time Series Forecasting via TARFLOW based VAE [3.7770105485163206]
  This paper presents TARFVAE, a novel generative framework that combines the Transformer-based autoregressive flow (TARFLOW) and variational autoencoder (VAE) for efficient one-step generative time series forecasting.<n>With simple modules, TARFVAE achieves superior performance over state-of-the-art deterministic and generative models across different forecast horizons on datasets.
  arXiv  Detail & Related papers  (2025-11-28T03:19:14Z)
• Advancing Reliable Test-Time Adaptation of Vision-Language Models under Visual Variations [67.35596444651037]
  Vision-language models (VLMs) exhibit remarkable zero-shot capabilities but struggle with distribution shifts in downstream tasks when labeled data is unavailable.<n>We propose a Reliable Test-time Adaptation (ReTA) method that enhances reliability from two perspectives.
  arXiv  Detail & Related papers  (2025-07-13T05:37:33Z)
• A Novel Framework for Significant Wave Height Prediction based on Adaptive Feature Extraction Time-Frequency Network [5.7146098061920885]
  Adaptive Feature Extraction Time-Frequency Network (AFE-TFNet) is proposed to improve prediction accuracy and stability.<n>It is an encoder-decoder rolling framework with two stages: feature extraction and feature fusion.<n>Results show AFE-TFNet significantly outperforms benchmark methods in terms of prediction accuracy.
  arXiv  Detail & Related papers  (2025-05-10T16:25:31Z)
• Delayed Feedback Modeling with Influence Functions [10.327472992234808]
  A major challenge is delayed feedback, where conversions may occur long after user interactions, leading to incomplete recent data and biased model training.<n>Existing solutions partially mitigate this issue but often rely on auxiliary models, making them computationally inefficient and less adaptive to user interest shifts.<n>We propose IF-DFM, an underlineInfluence underlineFunction-empowered for underlineDelayed underlineFeedback underlineModeling which estimates the impact of newly arrived and delayed conversions on model parameters, enabling efficient
  arXiv  Detail & Related papers  (2025-02-01T16:23:13Z)
• Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching [56.286064975443026]
  We make an interesting and somehow surprising observation: the computation of a large proportion of layers in the diffusion transformer, through a caching mechanism, can be readily removed even without updating the model parameters. We introduce a novel scheme, named Learningto-Cache (L2C), that learns to conduct caching in a dynamic manner for diffusion transformers. Experimental results show that L2C largely outperforms samplers such as DDIM and DPM-r, alongside prior cache-based methods at the same inference speed.
  arXiv  Detail & Related papers  (2024-06-03T18:49:57Z)
• Perceiver-based CDF Modeling for Time Series Forecasting [25.26713741799865]
  We propose a new architecture, called perceiver-CDF, for modeling cumulative distribution functions (CDF) of time series data. Our approach combines the perceiver architecture with a copula-based attention mechanism tailored for multimodal time series prediction. Experiments on the unimodal and multimodal benchmarks consistently demonstrate a 20% improvement over state-of-the-art methods.
  arXiv  Detail & Related papers  (2023-10-03T01:13:17Z)
• CARD: Channel Aligned Robust Blend Transformer for Time Series Forecasting [50.23240107430597]
  We design a special Transformer, i.e., Channel Aligned Robust Blend Transformer (CARD for short), that addresses key shortcomings of CI type Transformer in time series forecasting. First, CARD introduces a channel-aligned attention structure that allows it to capture both temporal correlations among signals. Second, in order to efficiently utilize the multi-scale knowledge, we design a token blend module to generate tokens with different resolutions. Third, we introduce a robust loss function for time series forecasting to alleviate the potential overfitting issue.
  arXiv  Detail & Related papers  (2023-05-20T05:16:31Z)
• RF+clust for Leave-One-Problem-Out Performance Prediction [0.9281671380673306]
  We study leave-one-problem-out (LOPO) performance prediction. We analyze whether standard random forest (RF) model predictions can be improved by calibrating them with a weighted average of performance values.
  arXiv  Detail & Related papers  (2023-01-23T16:14:59Z)
• Towards Long-Term Time-Series Forecasting: Feature, Pattern, and Distribution [57.71199089609161]
  Long-term time-series forecasting (LTTF) has become a pressing demand in many applications, such as wind power supply planning. Transformer models have been adopted to deliver high prediction capacity because of the high computational self-attention mechanism. We propose an efficient Transformerbased model, named Conformer, which differentiates itself from existing methods for LTTF in three aspects.
  arXiv  Detail & Related papers  (2023-01-05T13:59:29Z)
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting [25.417560221400347]
  Long sequence time-series forecasting (LSTF) demands a high prediction capacity. Recent studies have shown the potential of Transformer to increase the prediction capacity. We design an efficient transformer-based model for LSTF, named Informer, with three distinctive characteristics.
  arXiv  Detail & Related papers  (2020-12-14T11:43:09Z)
• Transformer Hawkes Process [79.16290557505211]
  We propose a Transformer Hawkes Process (THP) model, which leverages the self-attention mechanism to capture long-term dependencies. THP outperforms existing models in terms of both likelihood and event prediction accuracy by a notable margin. We provide a concrete example, where THP achieves improved prediction performance for learning multiple point processes when incorporating their relational information.
  arXiv  Detail & Related papers  (2020-02-21T13:48:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.