V.I.P. : Iterative Online Preference Distillation for Efficient Video Diffusion Models

• URL: http://arxiv.org/abs/2508.03254v1
• Date: Tue, 05 Aug 2025 09:31:54 GMT
• Title: V.I.P. : Iterative Online Preference Distillation for Efficient Video Diffusion Models
• Authors: Jisoo Kim, Wooseok Seo, Junwan Kim, Seungho Park, Sooyeon Park, Youngjae Yu,
• Abstract summary: We propose an effective distillation method, ReDPO, that integrates DPO and SFT.<n>Our approach leverages DPO to guide the student model to focus on recovering only the targeted properties, rather than passively imitating the teacher.<n>We additionally propose V.I.P., a novel framework for filtering and curating high-quality pair datasets, along with a step-by-step online approach for calibrated training.
• Score: 14.301804388786469
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: With growing interest in deploying text-to-video (T2V) models in resource-constrained environments, reducing their high computational cost has become crucial, leading to extensive research on pruning and knowledge distillation methods while maintaining performance. However, existing distillation methods primarily rely on supervised fine-tuning (SFT), which often leads to mode collapse as pruned models with reduced capacity fail to directly match the teacher's outputs, ultimately resulting in degraded quality. To address this challenge, we propose an effective distillation method, ReDPO, that integrates DPO and SFT. Our approach leverages DPO to guide the student model to focus on recovering only the targeted properties, rather than passively imitating the teacher, while also utilizing SFT to enhance overall performance. We additionally propose V.I.P., a novel framework for filtering and curating high-quality pair datasets, along with a step-by-step online approach for calibrated training. We validate our method on two leading T2V models, VideoCrafter2 and AnimateDiff, achieving parameter reduction of 36.2% and 67.5% each, while maintaining or even surpassing the performance of full models. Further experiments demonstrate the effectiveness of both ReDPO and V.I.P. framework in enabling efficient and high-quality video generation. Our code and videos are available at https://jiiiisoo.github.io/VIP.github.io/.

Related papers

• FEDS: Feature and Entropy-Based Distillation Strategy for Efficient Learned Image Compression [12.280695635625737]
  Learned image compression (LIC) methods have recently outperformed traditional codecs such as VVC in rate-distortion performance.<n>In this paper, we first construct a high-capacity teacher model by integrating Swin-Transformer V2-based attention modules.<n>We then propose a underlineFeature and underlineEntropy-based underlineDistillation underlineStrategy (textbfFEDS) that transfers key knowledge from the teacher to a lightweight student model.
  arXiv  Detail & Related papers  (2025-03-09T02:39:39Z)
• OnlineVPO: Align Video Diffusion Model with Online Video-Centric Preference Optimization [30.6130504613716]
  We introduce OnlineVPO, a preference learning approach tailored specifically for video diffusion models.<n>By employing the video reward model to offer concise video feedback on the fly, OnlineVPO offers effective and efficient preference guidance.
  arXiv  Detail & Related papers  (2024-12-19T18:34:50Z)
• From Slow Bidirectional to Fast Autoregressive Video Diffusion Models [52.32078428442281]
  Current video diffusion models achieve impressive generation quality but struggle in interactive applications due to bidirectional attention dependencies.<n>We address this limitation by adapting a pretrained bidirectional diffusion transformer to an autoregressive transformer that generates frames on-the-fly.<n>Our model achieves a total score of 84.27 on the VBench-Long benchmark, surpassing all previous video generation models.
  arXiv  Detail & Related papers  (2024-12-10T18:59:50Z)
• T2V-Turbo-v2: Enhancing Video Generation Model Post-Training through Data, Reward, and Conditional Guidance Design [79.7289790249621]
  Our proposed method, T2V-Turbo-v2, introduces a significant advancement by integrating various supervision signals. We highlight the crucial importance of tailoring datasets to specific learning objectives. We demonstrate the potential of this approach by extracting motion guidance from the training datasets and incorporating it into the ODE solver.
  arXiv  Detail & Related papers  (2024-10-08T04:30:06Z)
• Unleashing the Power of One-Step Diffusion based Image Super-Resolution via a Large-Scale Diffusion Discriminator [81.81748032199813]
  Diffusion models have demonstrated excellent performance for real-world image super-resolution (Real-ISR)<n>We propose a new One-Step textbfDiffusion model with a larger-scale textbfDiscriminator for SR.<n>Our discriminator is able to distill noisy features from any time step of diffusion models in the latent space.
  arXiv  Detail & Related papers  (2024-10-05T16:41:36Z)
• MiniVLN: Efficient Vision-and-Language Navigation by Progressive Knowledge Distillation [17.27883003990266]
  Vision-and-Language Navigation (VLN) is a core task in Embodied AI. This paper introduces a two-stage knowledge distillation framework, producing a student model, MiniVLN. Our findings indicate that the two-stage distillation approach is more effective in narrowing the performance gap between the teacher model and the student model.
  arXiv  Detail & Related papers  (2024-09-27T14:54:54Z)
• OSV: One Step is Enough for High-Quality Image to Video Generation [44.09826880566572]
  We introduce a two-stage training framework that effectively combines consistency distillation and GAN training.<n>We also propose a novel video discriminator design, which eliminates the need for decoding the video latents.<n>Our model is capable of producing high-quality videos in merely one-step, with the flexibility to perform multi-step refinement.
  arXiv  Detail & Related papers  (2024-09-17T17:16:37Z)
• Unsupervised Domain Adaption Harnessing Vision-Language Pre-training [4.327763441385371]
  This paper focuses on harnessing the power of Vision-Language Pre-training models in Unsupervised Domain Adaptation (UDA) We propose a novel method called Cross-Modal Knowledge Distillation (CMKD) Our proposed method outperforms existing techniques on standard benchmarks.
  arXiv  Detail & Related papers  (2024-08-05T02:37:59Z)
• One-Step Diffusion Distillation via Deep Equilibrium Models [64.11782639697883]
  We introduce a simple yet effective means of distilling diffusion models directly from initial noise to the resulting image. Our method enables fully offline training with just noise/image pairs from the diffusion model. We demonstrate that the DEQ architecture is crucial to this capability, as GET matches a $5times$ larger ViT in terms of FID scores.
  arXiv  Detail & Related papers  (2023-12-12T07:28:40Z)
• Parameter and Computation Efficient Transfer Learning for Vision-Language Pre-trained Models [79.34513906324727]
  In this paper, we aim at parameter and efficient transfer learning (PCETL) for vision-language pre-trained models. We propose a novel dynamic architecture skipping (DAS) approach towards effective PCETL.
  arXiv  Detail & Related papers  (2023-09-04T09:34:33Z)
• BOOT: Data-free Distillation of Denoising Diffusion Models with Bootstrapping [64.54271680071373]
  Diffusion models have demonstrated excellent potential for generating diverse images. Knowledge distillation has been recently proposed as a remedy that can reduce the number of inference steps to one or a few. We present a novel technique called BOOT, that overcomes limitations with an efficient data-free distillation algorithm.
  arXiv  Detail & Related papers  (2023-06-08T20:30:55Z)
• Dynamic Contrastive Distillation for Image-Text Retrieval [90.05345397400144]
  We present a novel plug-in dynamic contrastive distillation (DCD) framework to compress image-text retrieval models. We successfully apply our proposed DCD strategy to two state-of-the-art vision-language pretrained models, i.e. ViLT and METER. Experiments on MS-COCO and Flickr30K benchmarks show the effectiveness and efficiency of our DCD framework.
  arXiv  Detail & Related papers  (2022-07-04T14:08:59Z)

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.