Abstract:Diffusion Transformers (DiT) have emerged as powerful generative models for various tasks, including image, video, and speech synthesis. However, their inference process remains computationally expensive due to the repeated evaluation of resource-intensive attention and feed-forward modules. To address this, we introduce SmoothCache, a model-agnostic inference acceleration technique for DiT architectures. SmoothCache leverages the observed high similarity between layer outputs across adjacent diffusion timesteps. By analyzing layer-wise representation errors from a small calibration set, SmoothCache adaptively caches and reuses key features during inference. Our experiments demonstrate that SmoothCache achieves 8% to 71% speed up while maintaining or even improving generation quality across diverse modalities. We showcase its effectiveness on DiT-XL for image generation, Open-Sora for text-to-video, and Stable Audio Open for text-to-audio, highlighting its potential to enable real-time applications and broaden the accessibility of powerful DiT models.
Abstract:Many creative ideas are being proposed for image sensor designs, and these may be useful in applications ranging from consumer photography to computer vision. To understand and evaluate each new design, we must create a corresponding image processing pipeline that transforms the sensor data into a form that is appropriate for the application. The need to design and optimize these pipelines is time-consuming and costly. We explain a method that combines machine learning and image systems simulation that automates the pipeline design. The approach is based on a new way of thinking of the image processing pipeline as a large collection of local linear filters. We illustrate how the method has been used to design pipelines for novel sensor architectures in consumer photography applications.