HarmoniCa: Harmonizing Training and Inference for Better Feature Cache in Diffusion Transformer Acceleration

• URL: http://arxiv.org/abs/2410.01723v2
• Date: Fri, 4 Oct 2024 10:14:17 GMT
• Title: HarmoniCa: Harmonizing Training and Inference for Better Feature Cache in Diffusion Transformer Acceleration
• Authors: Yushi Huang, Zining Wang, Ruihao Gong, Jing Liu, Xinjie Zhang, Jun Zhang,
• Abstract summary: We propose a novel method that Harmonizes training and inference with a novel learning-based Caching framework. Compared to the traditional training paradigm, the newly proposed SDT maintains the continuity of the denoising process. IEPO integrates an efficient proxy mechanism to approximate the final image error caused by reusing the cached feature.
• Score: 18.170285241800798
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Diffusion Transformers (DiTs) have gained prominence for outstanding scalability and extraordinary performance in generative tasks. However, their considerable inference costs impede practical deployment. The feature cache mechanism, which involves storing and retrieving redundant computations across timesteps, holds promise for reducing per-step inference time in diffusion models. Most existing caching methods for DiT are manually designed. Although the learning-based approach attempts to optimize strategies adaptively, it suffers from discrepancies between training and inference, which hampers both the performance and acceleration ratio. Upon detailed analysis, we pinpoint that these discrepancies primarily stem from two aspects: (1) Prior Timestep Disregard, where training ignores the effect of cache usage at earlier timesteps, and (2) Objective Mismatch, where the training target (align predicted noise in each timestep) deviates from the goal of inference (generate the high-quality image). To alleviate these discrepancies, we propose HarmoniCa, a novel method that Harmonizes training and inference with a novel learning-based Caching framework built upon Step-Wise Denoising Training (SDT) and Image Error Proxy-Guided Objective (IEPO). Compared to the traditional training paradigm, the newly proposed SDT maintains the continuity of the denoising process, enabling the model to leverage information from prior timesteps during training, similar to the way it operates during inference. Furthermore, we design IEPO, which integrates an efficient proxy mechanism to approximate the final image error caused by reusing the cached feature. Therefore, IEPO helps balance final image quality and cache utilization, resolving the issue of training that only considers the impact of cache usage on the predicted output at each timestep.

Related papers

• ExpertFlow: Optimized Expert Activation and Token Allocation for Efficient Mixture-of-Experts Inference [41.41316718220569]
  ExpertFlow is designed to enhance inference efficiency by accommodating flexible routing and enabling efficient expert scheduling between CPU and GPU. Our experiments demonstrate that ExpertFlow achieves up to 93.72% GPU memory savings and enhances inference speed by 2 to 10 times compared to baseline methods.
  arXiv  Detail & Related papers  (2024-10-23T15:24:54Z)
• Temporal Feature Matters: A Framework for Diffusion Model Quantization [105.3033493564844]
  Diffusion models rely on the time-step for the multi-round denoising. We introduce a novel quantization framework that includes three strategies. This framework preserves most of the temporal information and ensures high-quality end-to-end generation.
  arXiv  Detail & Related papers  (2024-07-28T17:46:15Z)
• Adaptive Rentention & Correction for Continual Learning [114.5656325514408]
  A common problem in continual learning is the classification layer's bias towards the most recent task. We name our approach Adaptive Retention & Correction (ARC) ARC achieves an average performance increase of 2.7% and 2.6% on the CIFAR-100 and Imagenet-R datasets.
  arXiv  Detail & Related papers  (2024-05-23T08:43:09Z)
• Retraining-free Model Quantization via One-Shot Weight-Coupling Learning [41.299675080384]
  Mixed-precision quantization (MPQ) is advocated to compress the model effectively by allocating heterogeneous bit-width for layers. MPQ is typically organized into a searching-retraining two-stage process. In this paper, we devise a one-shot training-searching paradigm for mixed-precision model compression.
  arXiv  Detail & Related papers  (2024-01-03T05:26:57Z)
• On the Effectiveness of LayerNorm Tuning for Continual Learning in Vision Transformers [47.77328392236625]
  State-of-the-art rehearsal-free continual learning methods exploit the peculiarities of Vision Transformers to learn task-specific prompts. We introduce a two-stage training procedure, where we first optimize the task-specific parameters and then train the classifier with the same selection procedure of the inference time. Our method achieves results that are either superior or on par with the state of the art while being computationally cheaper.
  arXiv  Detail & Related papers  (2023-08-18T15:11:16Z)
• Adaptive Cross Batch Normalization for Metric Learning [75.91093210956116]
  Metric learning is a fundamental problem in computer vision. We show that it is equally important to ensure that the accumulated embeddings are up to date. In particular, it is necessary to circumvent the representational drift between the accumulated embeddings and the feature embeddings at the current training iteration.
  arXiv  Detail & Related papers  (2023-03-30T03:22:52Z)
• Accelerating Deep Learning Classification with Error-controlled Approximate-key Caching [72.50506500576746]
  We propose a novel caching paradigm, that we named approximate-key caching. While approximate cache hits alleviate DL inference workload and increase the system throughput, they however introduce an approximation error. We analytically model our caching system performance for classic LRU and ideal caches, we perform a trace-driven evaluation of the expected performance, and we compare the benefits of our proposed approach with the state-of-the-art similarity caching.
  arXiv  Detail & Related papers  (2021-12-13T13:49:11Z)
• Extrapolation for Large-batch Training in Deep Learning [72.61259487233214]
  We show that a host of variations can be covered in a unified framework that we propose. We prove the convergence of this novel scheme and rigorously evaluate its empirical performance on ResNet, LSTM, and Transformer.
  arXiv  Detail & Related papers  (2020-06-10T08:22:41Z)
• BERT Loses Patience: Fast and Robust Inference with Early Exit [91.26199404912019]
  We propose Patience-based Early Exit as a plug-and-play technique to improve the efficiency and robustness of a pretrained language model. Our approach improves inference efficiency as it allows the model to make a prediction with fewer layers.
  arXiv  Detail & Related papers  (2020-06-07T13:38:32Z)

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.