Inversion-DPO: Precise and Efficient Post-Training for Diffusion Models

• URL: http://arxiv.org/abs/2507.11554v4
• Date: Sun, 03 Aug 2025 03:07:39 GMT
• Title: Inversion-DPO: Precise and Efficient Post-Training for Diffusion Models
• Authors: Zejian Li, Yize Li, Chenye Meng, Zhongni Liu, Yang Ling, Shengyuan Zhang, Guang Yang, Changyuan Yang, Zhiyuan Yang, Lingyun Sun,
• Abstract summary: Inversion-DPO is an alignment framework that circumvents reward modeling.<n>Our method conducts intractable posterior sampling in Diffusion-DPO with the deterministic inversion from winning and losing samples to noise.<n>We apply Inversion-DPO to a basic task of text-to-image generation and a challenging task of compositional image generation.
• Score: 14.976130146925724
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent advancements in diffusion models (DMs) have been propelled by alignment methods that post-train models to better conform to human preferences. However, these approaches typically require computation-intensive training of a base model and a reward model, which not only incurs substantial computational overhead but may also compromise model accuracy and training efficiency. To address these limitations, we propose Inversion-DPO, a novel alignment framework that circumvents reward modeling by reformulating Direct Preference Optimization (DPO) with DDIM inversion for DMs. Our method conducts intractable posterior sampling in Diffusion-DPO with the deterministic inversion from winning and losing samples to noise and thus derive a new post-training paradigm. This paradigm eliminates the need for auxiliary reward models or inaccurate appromixation, significantly enhancing both precision and efficiency of training. We apply Inversion-DPO to a basic task of text-to-image generation and a challenging task of compositional image generation. Extensive experiments show substantial performance improvements achieved by Inversion-DPO compared to existing post-training methods and highlight the ability of the trained generative models to generate high-fidelity compositionally coherent images. For the post-training of compostitional image geneation, we curate a paired dataset consisting of 11,140 images with complex structural annotations and comprehensive scores, designed to enhance the compositional capabilities of generative models. Inversion-DPO explores a new avenue for efficient, high-precision alignment in diffusion models, advancing their applicability to complex realistic generation tasks. Our code is available at https://github.com/MIGHTYEZ/Inversion-DPO

Related papers

• LD-RPS: Zero-Shot Unified Image Restoration via Latent Diffusion Recurrent Posterior Sampling [23.886038479359918]
  We propose a dataset-free, and unified approach through recurrent posterior sampling utilizing a pretrained latent diffusion model.<n>Our method incorporates the multimodal understanding model to provide sematic priors for the generative model under a task-blind condition.
  arXiv  Detail & Related papers  (2025-07-01T14:25:09Z)
• Tuning Timestep-Distilled Diffusion Model Using Pairwise Sample Optimization [97.35427957922714]
  We present an algorithm named pairwise sample optimization (PSO), which enables the direct fine-tuning of an arbitrary timestep-distilled diffusion model.<n>PSO introduces additional reference images sampled from the current time-step distilled model, and increases the relative likelihood margin between the training images and reference images.<n>We show that PSO can directly adapt distilled models to human-preferred generation with both offline and online-generated pairwise preference image data.
  arXiv  Detail & Related papers  (2024-10-04T07:05:16Z)
• Pruning then Reweighting: Towards Data-Efficient Training of Diffusion Models [33.09663675904689]
  We investigate efficient diffusion training from the perspective of dataset pruning. Inspired by the principles of data-efficient training for generative models such as generative adversarial networks (GANs), we first extend the data selection scheme used in GANs to DM training. To further improve the generation performance, we employ a class-wise reweighting approach.
  arXiv  Detail & Related papers  (2024-09-27T20:21:19Z)
• Learning Energy-Based Models by Cooperative Diffusion Recovery Likelihood [64.95663299945171]
  Training energy-based models (EBMs) on high-dimensional data can be both challenging and time-consuming. There exists a noticeable gap in sample quality between EBMs and other generative frameworks like GANs and diffusion models. We propose cooperative diffusion recovery likelihood (CDRL), an effective approach to tractably learn and sample from a series of EBMs.
  arXiv  Detail & Related papers  (2023-09-10T22:05:24Z)
• Steerable Conditional Diffusion for Out-of-Distribution Adaptation in Medical Image Reconstruction [75.91471250967703]
  We introduce a novel sampling framework called Steerable Conditional Diffusion.<n>This framework adapts the diffusion model, concurrently with image reconstruction, based solely on the information provided by the available measurement.<n>We achieve substantial enhancements in out-of-distribution performance across diverse imaging modalities.
  arXiv  Detail & Related papers  (2023-08-28T08:47:06Z)
• Training Diffusion Models with Reinforcement Learning [82.29328477109826]
  Diffusion models are trained with an approximation to the log-likelihood objective. In this paper, we investigate reinforcement learning methods for directly optimizing diffusion models for downstream objectives. We describe how posing denoising as a multi-step decision-making problem enables a class of policy gradient algorithms.
  arXiv  Detail & Related papers  (2023-05-22T17:57:41Z)
• Hierarchical Integration Diffusion Model for Realistic Image Deblurring [71.76410266003917]
  Diffusion models (DMs) have been introduced in image deblurring and exhibited promising performance. We propose the Hierarchical Integration Diffusion Model (HI-Diff), for realistic image deblurring. Experiments on synthetic and real-world blur datasets demonstrate that our HI-Diff outperforms state-of-the-art methods.
  arXiv  Detail & Related papers  (2023-05-22T12:18:20Z)
• Restoration based Generative Models [0.886014926770622]
  Denoising diffusion models (DDMs) have attracted increasing attention by showing impressive synthesis quality. In this paper, we establish the interpretation of DDMs in terms of image restoration (IR) We propose a multi-scale training, which improves the performance compared to the diffusion process, by taking advantage of the flexibility of the forward process. We believe that our framework paves the way for designing a new type of flexible general generative model.
  arXiv  Detail & Related papers  (2023-02-20T00:53:33Z)
• Distribution-Aware Single-Stage Models for Multi-Person 3D Pose Estimation [29.430404703883084]
  We present a novel Distribution-Aware Single-stage (DAS) model for tackling the challenging multi-person 3D pose estimation problem. The proposed DAS model simultaneously localizes person positions and their corresponding body joints in the 3D camera space in a one-pass manner. Comprehensive experiments on benchmarks CMU Panoptic and MuPoTS-3D demonstrate the superior efficiency of the proposed DAS model.
  arXiv  Detail & Related papers  (2022-03-15T07:30:27Z)
• A Generic Approach for Enhancing GANs by Regularized Latent Optimization [79.00740660219256]
  We introduce a generic framework called em generative-model inference that is capable of enhancing pre-trained GANs effectively and seamlessly. Our basic idea is to efficiently infer the optimal latent distribution for the given requirements using Wasserstein gradient flow techniques.
  arXiv  Detail & Related papers  (2021-12-07T05:22:50Z)

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.