Diffusion Model Fine-Tuning

Ask about this topic

Overview

Diffusion model fine-tuning studies how to adapt pretrained image, video, audio, molecular, and language generators without rebuilding them from scratch. Parameter-efficient methods such as LoRA, prompt tuning, and adapters support personalization and domain adaptation, while supervised training and self-generated trajectories target tasks such as image editing, restoration, and masked-language-model refinement. For preference alignment and constrained generation, reinforcement learning methods such as GRPO, reward matching, and distillation optimize outputs using reward signals, often on benchmarks for text-image alignment, generation quality, or downstream utility. A central tension is improving the requested behavior while preserving the model’s general capabilities and diversity, with training cost, limited data, and reward hacking complicating that balance.

Latest papers

CardsList

181 papers32 papers in the last four weeks, up 100% on the four weeks before. 0.2% of all new papers.

Mar 13, 2026

Accelerating Stroke MRI with Diffusion Probabilistic Models through Large-Scale Pre-training and Target-Specific Fine-Tuning

Purpose: To develop a data-efficient strategy for accelerated MRI reconstruction with Diffusion Probabilistic Generative Models (DPMs) that enables faster scan times in clinical stroke MRI when only limited fully-sampled data samples are available. Methods: Our simple training strategy, inspired by the foundation model paradigm, first trains a DPM on a large, diverse collection of publicly available brain MRI data in fastMRI and then fine-tunes on a small dataset from the target application using carefully selected learning rates and fine-tuning durations. The approach is evaluated on controlled fastMRI experiments and on clinical stroke MRI data with a blinded clinical reader study. Results: DPMs pre-trained on approximately 4000 subjects with non-FLAIR contrasts and fine-tuned on FLAIR data from only 20 target subjects achieve reconstruction performance comparable to models trained with substantially more target-domain FLAIR data across multiple acceleration factors. Experiments reveal that moderate fine-tuning with a reduced learning rate yields improved performance, while insufficient or excessive fine-tuning degrades reconstruction quality. When applied to clinical stroke MRI, a blinded reader study involving two neuroradiologists indicates that images reconstructed using the proposed approach from 2×2 \times accelerated data are non-inferior to standard-of-care in terms of image quality and structural delineation. Conclusion: Large-scale pre-training combined with targeted fine-tuning enables DPM-based MRI reconstruction in data-constrained, accelerated clinical stroke MRI. The proposed approach substantially reduces the need for large application-specific datasets while maintaining clinically acceptable image quality, supporting the use of foundation-inspired diffusion models for accelerated MRI in targeted applications.

Yamin Arefeen, Sidharth Kumar, Steven Warach +2

Chandra Family Department of Electrical and Computer Engineering The University of Texas at Austin Austin, TX, USA · Department of Imaging Physics MD Anderson Cancer Center Houston, TX, USA · Dell Medical School Department of Neurology The University of Texas at Austin Austin, TX, USA +3

Accelerated MRI ReconstructionMagnetic Resonance ImagingDiffusion Model Fine-TuningDiffusion ModelsNeuroimaging
AskPDF
Mar 11, 2026

Continuous Diffusion Transformers for Designing Synthetic Regulatory Elements

We present a parameter-efficient Diffusion Transformer (DiT) for generating 200bp cell-type-specific regulatory DNA sequences. By replacing the U-Net backbone of DNA-Diffusion with a transformer denoiser equipped with a 2D CNN input encoder, our model matches the U-Net's best validation loss in 13 epochs (60×\times fewer) and converges 39% lower, while reducing memorization from 5.3% to 1.7% of generated sequences aligning to training data via BLAT. Ablations show the CNN encoder is essential: without it, validation loss increases 70% regardless of positional embedding choice. We further apply DDPO finetuning using Enformer as a reward model, achieving a 38×\times improvement in predicted regulatory activity. Cross-validation against DRAKES on an independent prediction task confirms that improvements reflect genuine regulatory signal rather than reward model overfitting.

Jonathan Liu, Kia Ghods

Department of Computer Science Princeton University

Molecular GenerationDiffusion TransformerConditional Diffusion ModelsDiffusion Model Fine-Tuning
AskPDF
Mar 11, 2026

Visually-Guided Controllable Medical Image Generation via Fine-Grained Semantic Disentanglement

Medical image synthesis is crucial for alleviating data scarcity and privacy constraints. However, fine-tuning general text-to-image (T2I) models remains challenging, mainly due to the significant modality gap between complex visual details and abstract clinical text. In addition, semantic entanglement persists, where coarse-grained text embeddings blur the boundary between anatomical structures and imaging styles, thus weakening controllability during generation. To address this, we propose a Visually-Guided Text Disentanglement framework. We introduce a cross-modal latent alignment mechanism that leverages visual priors to explicitly disentangle unstructured text into independent semantic representations. Subsequently, a Hybrid Feature Fusion Module (HFFM) injects these features into a Diffusion Transformer (DiT) through separated channels, enabling fine-grained structural control. Experimental results in three datasets demonstrate that our method outperforms existing approaches in terms of generation quality and significantly improves performance on downstream classification tasks. The source code is available at https://github.com/hx111/VG-MedGen.

Xin Huang, Junjie Liang, Qingshan Hou +4

Computer Science and Engineering, Northeastern University, Shenyang, China · Key Laboratory of Intelligent Computing in Medical Image of Ministry of Education, Northeastern University, Shenyang, China · National Frontiers Science Center for Industrial Intelligence and Systems Optimization, Shenyang, China +2

Medical Image SynthesisControllable Image GenerationT2I Diffusion ModelsMultimodal Disentangled Representation LearningDiffusion Model Fine-Tuning
AskPDF
Mar 10, 2026

BLK-Assist: A Methodological Framework for Artist-Led Co-Creation with Generative AI Models

This paper presents BLK-Assist, a modular framework for artist-specific fine-tuning of diffusion models using parameter-efficient methods. The system is implemented as a case study with a single professional artist's proprietary corpus and consists of three components: BLK-Conceptor (LoRA-adapted conceptual sketch generation), BLK-Stencil (LayerDiffuse-based transparency-preserving asset generation), and BLK-Upscale (hybrid Real-ESRGAN and texture-conditioned diffusion for high-resolution outputs). We document dataset composition, preprocessing, training configurations, and inference workflows to enable reproducibility with publicly available models to illustrate a privacy-preserving, consent-based approach to human-AI co-creation that maintains stylistic fidelity to the source corpus and can be adapted for other artists under similar constraints.

Daniel Grimes, Rachel M. Harrison

Visual-Synthesizer.com Grass Valley, CA 95945 · Ophiuchus LLC Dover, DE 19904

Image GenerationHuman-AI Co-CreationDiffusion Model Fine-Tuning
AskPDF
Mar 9, 2026

Diffusion-Based Data Augmentation for Image Recognition: A Systematic Analysis and Evaluation

Diffusion-based data augmentation (DiffDA) has emerged as a promising approach to improving classification performance under data scarcity. However, existing works vary significantly in task configurations, model choices, and experimental pipelines, making it difficult to fairly compare methods or assess their effectiveness across different scenarios. Moreover, there remains a lack of systematic understanding of the full DiffDA workflow. In this work, we introduce UniDiffDA, a unified analytical framework that decomposes DiffDA methods into three core components: model fine-tuning, sample generation, and sample utilization. This perspective enables us to identify key differences among existing methods and clarify the overall design space. Building on this framework, we develop a comprehensive and fair evaluation protocol, benchmarking representative DiffDA methods across diverse low-data classification tasks. Extensive experiments reveal the relative strengths and limitations of different DiffDA strategies and offer practical insights into method design and deployment. All methods are re-implemented within a unified codebase, with full release of code and configurations to ensure reproducibility and to facilitate future research.

Zekun Li, Yinghuan Shi, Yang Gao +1

State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China. · School of Computing and Data Science, The University of Hong Kong, Hong Kong SAR, China.

Diffusion-Based Data AugmentationData AugmentationDiffusion Model Fine-TuningFew-Shot Image Classification
AskPDF
Mar 8, 2026

PureCC: Pure Learning for Text-to-Image Concept Customization

Existing concept customization methods have achieved remarkable outcomes in high-fidelity and multi-concept customization. However, they often neglect the influence on the original model's behavior and capabilities when learning new personalized concepts. To address this issue, we propose PureCC. PureCC introduces a novel decoupled learning objective for concept customization, which combines the implicit guidance of the target concept with the original conditional prediction. This separated form enables PureCC to substantially focus on the original model during training. Moreover, based on this objective, PureCC designs a dual-branch training pipeline that includes a frozen extractor providing purified target concept representations as implicit guidance and a trainable flow model producing the original conditional prediction, jointly achieving pure learning for personalized concepts. Furthermore, PureCC introduces a novel adaptive guidance scale λ⋆λ^\star to dynamically adjust the guidance strength of the target concept, balancing customization fidelity and model preservation. Extensive experiments show that PureCC achieves state-of-the-art performance in preserving the original behavior and capabilities while enabling high-fidelity concept customization. The code is available at https://github.com/lzc-sg/PureCC.

Zhichao Liao, Xiaole Xian, Qingyu Li +7

Tsinghua University · School of Computer Science & Software Engineering, Shenzhen University · Kling Team, Kuaishou Technology +3

Subject-Driven Image GenerationDiffusion Model Fine-TuningConditional Generative Modeling
AskPDF
Mar 7, 2026

Diffusion Controller: Framework, Algorithms and Parameterization

Controllable diffusion generation often relies on various heuristics that are seemingly disconnected without a unified understanding. We bridge this gap with Diffusion Controller (DiffCon), a unified control-theoretic view that casts reverse diffusion sampling as state-only stochastic control within (generalized) linearly-solvable Markov Decision Processes (LS-MDPs). Under this framework, control acts by reweighting the pretrained reverse-time transition kernels, balancing terminal objectives against an ff-divergence cost. From the resulting optimality conditions, we derive practical reinforcement learning methods for diffusion fine-tuning: (i) f-divergence-regularized policy-gradient updates, including a PPO-style rule, and (ii) a regularizer-determined reward-weighted regression objective with a minimizer-preservation guarantee under the Kullback-Leibler (KL) divergence. The LS-MDP framework further implies a principled model form: the optimal score decomposes into a fixed pretrained baseline plus a lightweight control correction, motivating a side-network parameterization conditioned on exposed intermediate denoising outputs, enabling effective gray-box adaptation with a frozen backbone. Experiments on Stable Diffusion v1.4 across supervised and reward-driven finetuning show consistent gains in preference-alignment win rates and improved quality-efficiency trade-offs versus gray-box baselines and even the parameter-efficient white-box adapter LoRA.

Tong Yang, Moonkyung Ryu, Chih-Wei Hsu +4

Carnegie Mellon University; Intern at Google Research · Google · Yale University

Diffusion Model AdaptationDiffusion Model Fine-TuningDiffusion ModelsDiffusion Model AlignmentRL for Diffusion ModelsDiffusion-Based Denoising
AskPDF
Mar 6, 2026

Optimizing 3D Diffusion Models for Medical Imaging via Multi-Scale Reward Learning

Diffusion models have emerged as powerful tools for 3D medical image generation, yet bridging the gap between standard training objectives and clinical relevance remains a challenge. This paper presents a method to enhance 3D diffusion models using Reinforcement Learning (RL) with multi-scale feedback. We first pretrain a 3D diffusion model on MRI volumes to establish a robust generative prior. Subsequently, we fine-tune the model using Proximal Policy Optimization (PPO), guided by a novel reward system that integrates both 2D slice-wise assessments and 3D volumetric analysis. This combination allows the model to simultaneously optimize for local texture details and global structural coherence. We validate our framework on the BraTS 2019 and OASIS-1 datasets. Our results indicate that incorporating RL feedback effectively steers the generation process toward higher quality distributions. Quantitative analysis reveals significant improvements in Fréchet Inception Distance (FID) and, crucially, the synthetic data demonstrates enhanced utility in downstream tumor and disease classification tasks compared to non-optimized baselines.

Yueying Tian, Xudong Han, Meng Zhou +3

University of Sussex · University of Toronto

Medical Image Synthesis3D Diffusion Models3D Medical ImagingDiffusion Model Fine-TuningRL for Diffusion Models
AskPDF
Mar 4, 2026

Tuning Just Enough: Lightweight Backdoor Attacks on Multi-Encoder Diffusion Models

As text-to-image diffusion models become increasingly deployed in real-world applications, concerns about backdoor attacks have gained significant attention. Prior work on text-based backdoor attacks has largely focused on diffusion models conditioned on a single lightweight text encoder. However, more recent diffusion models that incorporate multiple large-scale text encoders remain underexplored in this context. Given the substantially increased number of trainable parameters introduced by multiple text encoders, an important question is whether backdoor attacks can remain both efficient and effective in such settings. In this work, we study Stable Diffusion 3, which uses three distinct text encoders and has not yet been systematically analyzed for text-encoder-based backdoor vulnerabilities. To understand the role of text encoders in backdoor attacks, we define four categories of attack targets and identify the minimal sets of encoders required to achieve effective performance for each attack objective. Based on this, we further propose Multi-Encoder Lightweight aTtacks (MELT), which trains only low-rank adapters while keeping the pretrained text encoder weight frozen. We demonstrate that tuning fewer than 0.2% of the total encoder parameters is sufficient for successful backdoor attacks on Stable Diffusion 3, revealing previously underexplored vulnerabilities in practical attack scenarios in multi-encoder settings.

Ziyuan Chen, Yujin Jeong, Tobias Braun +1

Computer Science Department, TU Darmstadt, Germany · Hessian Center for Artificial Intelligence (hessian.AI), Darmstadt, Germany

Backdoor AttacksT2I Diffusion ModelsDiffusion ModelsDiffusion Model Fine-Tuning
AskPDF
Mar 2, 2026

CRoCoDiL: Continuous and Robust Conditioned Diffusion for Language

Masked Diffusion Models (MDMs) provide an efficient non-causal alternative to autoregressive generation but often struggle with token dependencies and semantic incoherence due to their reliance on discrete marginal distributions. We address these limitations by shifting the diffusion process into a continuous sentence-level semantic space. We propose CRoCoDiL (Continuous and Robust Conditioned Diffusion for Language), a unified fine-tuning approach that jointly trains an encoder-demasker architecture, grounding the MDM demasking in continuous latent representations. This leads to the formation of a novel autoencoder in which decoding is obtained by an MDM algorithm. Relying on the same framework, we introduce two unconditional text synthesis algorithms: Continuous-Then-Discrete (ConThenDisc), a hybrid-diffusion approach that first generates latent representations in continuous space and then decodes these to tokens via an MDM, and Continuous-Within-Discrete (ConWithinDisc), a multi-diffusion strategy that refines latent representations throughout the discrete sampling process. Experiments using LLaDA show that our methods achieve superior generation quality and more than 10x faster sampling speeds in an unconditional setting.

Roy Uziel, Omer Belhasin, Itay Levy +4

NVIDIA

Diffusion Language ModelsContinuous Diffusion Language ModelsText GenerationMasked Diffusion ModelsDiffusion Model Fine-Tuning
AskPDF
Feb 23, 2026

ChimeraLoRA: Multi-Head LoRA-Guided Synthetic Datasets

Beyond general recognition tasks, specialized domains and fine-grained settings often encounter data scarcity, especially for tail classes. To obtain less biased and more reliable models under such scarcity, practitioners leverage diffusion models to supplement underrepresented regions of real data. Specifically, recent studies fine-tune pretrained diffusion models with LoRA on few-shot real sets to synthesize additional images. While an image-wise LoRA trained on a single image captures fine-grained details yet offers limited diversity, a class-wise LoRA trained over all shots produces diverse images as it encodes class priors yet tends to overlook fine details. To combine both benefits, we separate the adapter into a class-shared LoRA~AA for class priors and per-image LoRAs~B\mathcal{B} for image-specific characteristics. To expose coherent class semantics in the shared LoRA~AA, we propose a semantic boosting by preserving class bounding boxes during training. For generation, we compose AA with a mixture of B\mathcal{B} using coefficients drawn from a Dirichlet distribution. Across diverse datasets, our synthesized images are both diverse and detail-rich while closely aligning with the few-shot real distribution, yielding robust gains in downstream classification accuracy.

Hoyoung Kim, Minwoo Jang, Jabin Koo +2

Graduate School of AI, POSTECH · Dept. of CSE, POSTECH · NAVER AI Lab

Synthetic Data AugmentationDiffusion Model Fine-TuningLow-Rank AdaptationFew-Shot Image Classification
AskPDF
Feb 22, 2026

DGPO: RL-Steered Graph Diffusion for Neural Architecture Generation

Reinforcement learning fine-tuning has proven effective for steering generative diffusion models toward desired properties in image and molecular domains. Graph diffusion models have similarly been applied to combinatorial structure generation, including neural architecture search (NAS). However, neural architectures are directed acyclic graphs (DAGs) where edge direction encodes functional semantics such as data flow-information that existing graph diffusion methods, designed for undirected structures, discard. We propose Directed Graph Policy Optimization (DGPO), which extends reinforcement learning fine-tuning of discrete graph diffusion models to DAGs via topological node ordering and positional encoding. Validated on NAS-Bench-101 and NAS-Bench-201, DGPO matches the benchmark optimum on all three NAS-Bench-201 tasks (91.61%, 73.49%, 46.77%). The central finding is that the model learns transferable structural priors: pretrained on only 7% of the search space, it generates near-oracle architectures after fine-tuning, within 0.32 percentage points of the full-data model and extrapolating 7.3 percentage points beyond its training ceiling. Bidirectional control experiments confirm genuine reward-driven steering, with inverse optimization reaching near random-chance accuracy (9.5%). These results demonstrate that reinforcement learning-steered discrete diffusion, once extended to handle directionality, provides a controllable generative framework for directed combinatorial structures.

Aleksei Liuliakov, Luca Hermes, Barbara Hammer

Machine Learning Group, Bielefeld University, Germany

Graph GenerationGraph DiffusionNeural Architecture SearchDiffusion Model Fine-TuningRL for Diffusion Models
AskPDF
Feb 22, 2026

RegionRoute: Regional Style Transfer with Diffusion Model

Precise spatial control in diffusion-based style transfer remains challenging. This challenge arises because diffusion models treat style as a global feature and lack explicit spatial grounding of style representations, making it difficult to restrict style application to specific objects or regions. To our knowledge, existing diffusion models are unable to perform true localized style transfer, typically relying on handcrafted masks or multi-stage post-processing that introduce boundary artifacts and limit generalization. To address this, we propose an attention-supervised diffusion framework that explicitly teaches the model where to apply a given style by aligning the attention scores of style tokens with object masks during training. Two complementary objectives, a Focus loss based on KL divergence and a Cover loss using binary cross-entropy, jointly encourage accurate localization and dense coverage. A modular LoRA-MoE design further enables efficient and scalable multi-style adaptation. To evaluate localized stylization, we introduce the Regional Style Editing Score, which measures Regional Style Matching through CLIP-based similarity within the target region and Identity Preservation via masked LPIPS and pixel-level consistency on unedited areas. Experiments show that our method achieves mask-free, single-object style transfer at inference, producing regionally accurate and visually coherent results that outperform existing diffusion-based editing approaches.

Bowen Chen, Jake Zuena, Alan C. Bovik +1

The University of Texas at Austin · Dolby Laboratories, Inc.

Image Style TransferDiffusion-Based Image EditingAttention Control in Diffusion ModelsDiffusion Model Fine-Tuning
AskPDF
Feb 18, 2026

Efficient Tail-Aware Generative Optimization via Flow Model Fine-Tuning

Fine-tuning pre-trained diffusion and flow models to optimize downstream utilities is central to real-world deployment. Existing entropy-regularized methods primarily maximize expected reward, providing no mechanism to shape tail behavior. However, tail control is often essential: the lower tail determines reliability by limiting low-reward failures, while the upper tail enables discovery by prioritizing rare, high-reward outcomes. In this work, we present Tail-aware Flow Fine-Tuning (TFFT), a principled and efficient distributional fine-tuning algorithm based on the Conditional Value-at-Risk (CVaR). We address two distinct tail-shaping goals: right-CVaR for seeking novel samples in the high-reward tail and left-CVaR for controlling worst-case samples in the low-reward tail. Unlike prior approaches that rely on non-linear optimization, we leverage the variational dual formulation of CVaR to decompose it into a decoupled two-stage procedure: a lightweight one-dimensional threshold optimization step, and a single entropy-regularized fine-tuning process via a specific pseudo-reward. This decomposition achieves CVaR fine-tuning efficiently with computational cost comparable to standard expected fine-tuning methods. We demonstrate the effectiveness of TFFT across illustrative experiments, high-dimensional text-to-image generation, and molecular design.

Zifan Wang, Riccardo De Santi, Xiaoyu Mo +3

KTH Royal Institute of Technology · ETH Zurich · ETH AI Center +2

Molecule DesignFine-TuningDiffusion Model Fine-TuningStochastic Optimization
AskPDF
Feb 16, 2026

Efficient Text-Guided Convolutional Adapter for the Diffusion Model

We introduce the Nexus Adapters, novel text-guided efficient adapters to the diffusion-based framework for the Structure Preserving Conditional Generation (SPCG). Recently, structure-preserving methods have achieved promising results in conditional image generation by using a base model for prompt conditioning and an adapter for structure input, such as sketches or depth maps. These approaches are highly inefficient and sometimes require equal parameters in the adapter compared to the base architecture. It is not always possible to train the model since the diffusion model is itself costly, and doubling the parameter is highly inefficient. In these approaches, the adapter is not aware of the input prompt; therefore, it is optimal only for the structural input but not for the input prompt. To overcome the above challenges, we proposed two efficient adapters, Nexus Prime and Slim, which are guided by prompts and structural inputs. Each Nexus Block incorporates cross-attention mechanisms to enable rich multimodal conditioning. Therefore, the proposed adapter has a better understanding of the input prompt while preserving the structure. We conducted extensive experiments on the proposed models and demonstrated that the Nexus Prime adapter significantly enhances performance, requiring only 8M additional parameters compared to the baseline, T2I-Adapter. Furthermore, we also introduced a lightweight Nexus Slim adapter with 18M fewer parameters than the T2I-Adapter, which still achieved state-of-the-art results. Code: https://github.com/arya-domain/Nexus-Adapters

Aryan Das, Koushik Biswas, Swalpa Kumar Roy +2

VIT Bhopal · IIIT Delhi · Tezpur University Assam +1

Diffusion Model AdaptationConditional Image GenerationT2I Diffusion ModelsMultimodal Diffusion ModelsParameter-Efficient Fine-TuningDiffusion Model Fine-Tuning
AskPDF
Feb 13, 2026

SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning

Many training-free sparse attention methods are effective for accelerating diffusion models. Recently, several works suggest that making sparse attention trainable can further increase sparsity while preserving generation quality. We study three key questions: (1) when do the two common masking rules, i.e., Top-k and Top-p, fail, and how can we avoid these failures? (2) why can trainable sparse attention reach higher sparsity than training-free methods? (3) what are the limitations of fine-tuning sparse attention using the diffusion loss, and how can we address them? Based on this analysis, we propose SpargeAttention2, a trainable sparse attention method that achieves high sparsity without degrading generation quality. SpargeAttention2 includes (i) a hybrid masking rule that combines Top-k and Top-p for more robust masking at high sparsity, (ii) an efficient trainable sparse attention implementation, and (iii) a distillation-inspired fine-tuning objective to better preserve generation quality during fine-tuning using sparse attention. Experiments on video diffusion models show that SpargeAttention2 reaches 95% attention sparsity and a 16.2x attention speedup while maintaining generation quality, consistently outperforming prior sparse attention methods.

Jintao Zhang, Kai Jiang, Chendong Xiang +5

Tsinghua University · UC Berkeley

Sparse AttentionDiffusion Model AccelerationDiffusion Model Fine-TuningVideo Diffusion Models
AskPDF
Feb 8, 2026

Selective Fine-Tuning for Targeted and Robust Concept Unlearning

Text guided diffusion models are used by millions of users, but can be easily exploited to produce harmful content. Concept unlearning methods aim at reducing the models' likelihood of generating harmful content. Traditionally, this has been tackled at an individual concept level, with only a handful of recent works considering more realistic concept combinations. However, state of the art methods depend on full finetuning, which is computationally expensive. Concept localisation methods can facilitate selective finetuning, but existing techniques are static, resulting in suboptimal utility. In order to tackle these challenges, we propose TRUST (Targeted Robust Selective fine Tuning), a novel approach for dynamically estimating target concept neurons and unlearning them through selective finetuning, empowered by a Hessian based regularization. We show experimentally, against a number of SOTA baselines, that TRUST is robust against adversarial prompts, preserves generation quality to a significant degree, and is also significantly faster than the SOTA. Our method achieves unlearning of not only individual concepts but also combinations of concepts and conditional concepts, without any specific regularization.

Mansi, Avinash Kori, Francesca Toni +1

Department of Computing Imperial College London, United Kingdom

Fine-TuningAdversarial RobustnessDiffusion Model Fine-TuningDiffusion Model UnlearningSelective Fine-TuningConcept Erasure+1
AskPDF
Feb 4, 2026

Rethinking the Design Space of Reinforcement Learning for Diffusion Models: On the Importance of Likelihood Estimation Beyond Loss Design

Reinforcement learning has been widely applied to diffusion and flow models for visual tasks such as text-to-image generation. However, these tasks remain challenging because diffusion models have intractable likelihoods, which creates a barrier for directly applying popular policy-gradient type methods. Existing approaches primarily focus on crafting new objectives built on already heavily engineered LLM objectives, using ad hoc estimators for likelihood, without a thorough investigation into how such estimation affects overall algorithmic performance. In this work, we provide a systematic analysis of the RL design space by disentangling three factors: i) policy-gradient objectives, ii) likelihood estimators, and iii) rollout sampling schemes. We show that adopting an evidence lower bound (ELBO) based model likelihood estimator, computed only from the final generated sample, is the dominant factor enabling effective, efficient, and stable RL optimization, outweighing the impact of the specific policy-gradient loss functional. We validate our findings across multiple reward benchmarks using SD 3.5 Medium, and observe consistent trends across all tasks. Our method improves the GenEval score from 0.24 to 0.95 in 90 GPU hours, which is 4.6 times more efficient than FlowGRPO and 2×2\times more efficient than the SOTA method without reward hacking.

Jaemoo Choi, Yuchen Zhu, Wei Guo +6

Georgia Institute of Technology · National University of Singapore · Nanjing university.

T2I GenerationDiffusion Model Fine-TuningPolicy Gradient MethodsRL for Diffusion ModelsDiffusion Models
AskPDF
Jan 21, 2026

Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models

Post-training pretrained autoregressive models (ARMs) into masked diffusion models (MDMs) provides an efficient route to diffusion language modeling, but it remains unclear whether the resulting models reuse inherited autoregressive computation or reorganize it for non-autoregressive generation. We compare two 7B ARM-MDM families across four controlled diagnostic tasks and find a task-dependent mechanism shift. On prefix-dominant tasks, MDMs largely preserve inherited high-attribution pathways or exhibit only modest changes in where computation occurs. On globally constrained tasks, the reorganization is substantially stronger, with task-relevant computation shifting toward earlier layers. This depth-wise pattern persists across prompt resampling, circuit budgets, and tested inference budgets, while targeted ablations support the functional importance of the identified structures under the tested intervention protocols. At the component level, diagnostic probes suggest that ARMs rely more strongly on sharply specialized components, whereas MDMs exhibit weaker single-component specialization and more diffuse output-space alignment. Together, these results suggest that diffusion post-training selectively preserves or reorganizes inherited computation according to task structure, rather than uniformly replacing autoregressive mechanisms.

Injin Kong, Hyoungjoon Lee, Yohan Jo

Graduate School of Data Science, Seoul National University · Department of Biosystems & Biomaterials Science and Engineering, Seoul National University

Masked Diffusion ModelsAutoregressive DiffusionAutoregressive Language ModelingMechanistic InterpretabilityDiffusion Model Fine-TuningLanguage Modeling+2
AskPDF
Jan 18, 2026

Beyond the Dirac Delta: Mitigating Diversity Collapse in Reinforcement Fine-Tuning for Versatile Image Generation

Reinforcement learning (RL) has emerged as a paradigm for fine-tuning large-scale generative models, such as diffusion and flow models, to align with complex human preferences and user-specified tasks. A fundamental limitation remains the curse of diversity collapse, where the objective formulation and optimization landscape inherently collapse the policy to a Dirac delta distribution. To address this challenge, we propose DRIFT (DiveRsity-Incentivized Reinforcement Fine-Tuning for Versatile Image Generation), an innovative framework that systematically incentivizes output diversity throughout the on-policy fine-tuning process, reconciling strong task alignment with high generation diversity to enhance versatility essential for applications that demand diverse candidate generations. We approach the problem across three representative perspectives: i) sampling a reward-concentrated subset that filters out reward outliers to prevent premature collapse; ii) prompting with stochastic variations to expand the conditioning space, and iii) optimization of the intra-group diversity with a potential-based reward shaping mechanism. Experimental results show that DRIFT exhibits clear Pareto dominance in task alignment and generation diversity, achieving 3.88%∼\sim81.42% higher diversity at matched alignment and 14.38%∼\sim56.45% higher alignment at matched diversity. Our code is available at \textcolor{magenta}{https://github.com/NJU-RL/DRIFT}.

Jinmei Liu, Haoru Li, Zhenhong Sun +6

Nanjing University · Australian National University · Wuhan University +2

RL for Image GenerationBehavioral Diversity in RLDiffusion Model Fine-TuningDiversity-Aware Sampling
AskPDF
Dec 29, 2025

Direct Diffusion Score Preference Optimization via Stepwise Contrastive Policy-Pair Supervision

Diffusion models have achieved impressive results in generative tasks such as text-to-image synthesis, yet they often struggle to fully align outputs with nuanced user intent and maintain consistent aesthetic quality. Existing preference-based training methods such as Diffusion Direct Preference Optimization help address these issues, but obtain their supervision targets from the forward process q(xt−1 ⁣∣ ⁣xt,x0)q(\boldsymbol{x}_{t-1}\!\mid\!\boldsymbol{x}_t,\boldsymbol{x}_0) derived from terminal samples, which is not directly aligned with the model's actual backward denoising transitions at each step. In this work, we introduce Direct Diffusion Score Preference Optimization (DDSPO), which defines stepwise preference supervision directly over backward denoising transitions through a contrastive policy pair, rather than relying on forward-process approximations from terminal samples. We propose two practical instantiations of the contrastive policy pair: training separate winning and losing models on preference data, and inducing a contrastive policy pair without additional training by using a pretrained reference model conditioned on an original prompt and a semantically degraded variant, requiring neither reward modeling nor manual annotations. Empirical results show that contrastive-policy-pair supervision is more effective than forward-process-based supervision across text-image alignment and aesthetic-quality tasks. Our implementation is available at: https://dohyun-as.github.io/DDSPO

Dohyun Kim, Seungwoo Lyu, Seung Wook Kim +1

Dept. of CSE, Korea University · 2NVIDIA

Diffusion Model AlignmentDiffusion Model Fine-TuningPreference Optimization for Diffusion ModelsT2I Diffusion Models
AskPDF
Dec 22, 2025

MixFlow Training: Alleviating Exposure Bias with Slowed Interpolation Mixture

This paper studies the training-testing discrepancy (a.k.a. exposure bias) problem for improving the diffusion models. During training, the input of a prediction network at one training timestep is the corresponding ground-truth noisy data that is an interpolation of the noise and the data, and during testing, the input is the generated noisy data. We present a novel training approach, named MixFlow, for improving the performance. Our approach is motivated by the Slow Flow phenomenon: the ground-truth interpolation that is the nearest to the generated noisy data at a given sampling timestep is observed to correspond to a higher-noise timestep (termed slowed timestep), i.e., the corresponding ground-truth timestep is slower than the sampling timestep. MixFlow leverages the interpolations at the slowed timesteps, named slowed interpolation mixture, for post-training the prediction network for each training timestep. Experiments over class-conditional image generation (including SiT, REPA, and RAE) and text-to-image generation validate the effectiveness of our approach. Our approach MixFlow over the RAE models achieve strong generation results on ImageNet: 1.43 FID (without guidance) and 1.10 (with guidance) at 256 x 256, and 1.55 FID (without guidance) and 1.10 (with guidance) at 512 x 512.

Hui Li, Fu-Yun Wang, Haoyuan Xia +4

Shanghai Innovation Institute · Fudan University · Shanghai Academy of AI for Science +2

Diffusion Model Fine-TuningDiffusion-Based Image GenerationT2I Diffusion ModelsDiffusion Models
AskPDF
Dec 19, 2025

MatLat: Material Latent Space for PBR Texture Generation

We propose a generative framework for producing high-quality PBR textures on a given 3D mesh. As large-scale PBR texture datasets are scarce, our approach focuses on effectively leveraging the embedding space and diffusion priors of pretrained latent image generative models while learning a material latent space, MatLat, through targeted fine-tuning. Unlike prior methods that freeze the embedding network, which leads to distribution shifts when encoding additional PBR channels and hinders subsequent diffusion training, we fine-tune the pretrained VAE so that new material channels can be incorporated with minimal latent distribution deviation. We further show that correspondence-aware attention alone is insufficient for cross-view consistency unless the latent-to-image mapping preserves locality. To enforce this locality, we introduce a regularization in the VAE fine-tuning that crops latent patches, decodes them, and aligns the corresponding image regions to maintain strong pixel-latent spatial correspondence. Ablation studies and comparison with previous baselines demonstrate that our framework improves PBR texture fidelity and that each component is critical for achieving state-of-the-art performance.

Kyeongmin Yeo, Yunhong Min, Jaihoon Kim +1

KAIST

Texture SynthesisLatent Diffusion ModelsDiffusion Model Fine-Tuning
AskPDF
Nov 27, 2025

Anchor to Expand: Semantic Anchoring for Personalized Text-to-Image Diffusion Models

Personalizing text-to-image diffusion models extends pretrained models to represent novel user-specific concepts from only a few reference images. However, learning a new concept while building on the prior knowledge of the pretrained model remains a key challenge. When personalization focuses on learning the target concept, the model tends to overfit the reference examples and degrade its general capability. In contrast, emphasizing prior preservation can hinder capturing distinctive personalized attributes. In this paper, we address this challenge by viewing a personalized concept as an underrepresented concept whose semantic counterpart is well represented in the pretrained model. Rather than treating the learning of a new concept and prior preservation as separate objectives, we reformulate them as a single anchored learning problem. We therefore introduce Semantic Anchoring Personalization (SAP), which keeps concept learning grounded in the pretrained semantic structure while capturing subject-specific attributes. The proposed objective offers a simple yet effective formulation that can be applied across different model backbones without architectural modifications or auxiliary networks. Extensive experiments across various settings demonstrate that SAP achieves a better balance between subject fidelity and text-image alignment than baseline methods. Further ablation studies validate the contribution of semantic anchoring to personalization.

Seoyun Yang, Gihoon Kim, Taesup Kim

Graduate School of Data Science, Seoul National University

Subject-Driven Image GenerationT2I GenerationDiffusion Model Fine-TuningT2I Diffusion Models
AskPDF
Nov 24, 2025

Rethinking Garment Conditioning in Diffusion-based Virtual Try-On: Decouple, Don't Denoise

Virtual Try-On (VTON) synthesizes realistic images of a person wearing a target garment, with broad applications in e-commerce and fashion. Diffusion-based dual-UNet methods achieve strong results but double the parameters by dedicating a separate network to garment conditioning. Spatial concatenation offers a simpler single-network alternative, yet both UNet- and DiT-based instantiations report that full fine-tuning is ineffective, and the community has settled for attention-only training. We ask: why does full fine-tuning fail, and can this be resolved? Through what is, to our knowledge, the first visualization study of dual-UNet reference network behavior, we identify a unifying insight: garment conditioning must be decoupled from the denoising process. Spatial concatenation violates this by embedding the garment within the denoising target, causing three conflicts: guidance leakage, gradient competition, and train-test discrepancy. We derive three design principles to restore this decoupling and implement them as a pure recipe atop a standard architecture with no modification. The resulting model, DeCo-VTON (860M params), achieves single-network state of the art, matching the dual-UNet state of the art at half the cost while being preferred in human evaluation.

Kihyun Na, Jinyoung Choi, Injung Kim

Advanced AI Talent Education and Research Group for Industrial Innovation, Handong Global University, Pohang 37554, Republic of Korea · Department of Artificial Intelligence, Computer and Electrical Engineering, Handong Global University, Pohang 37554, Republic of Korea

Virtual Try-OnDiffusion Model Fine-TuningConditional Diffusion Models
AskPDF
Nov 22, 2025

FeRA: Frequency-Energy Constrained Routing for Effective Diffusion Adaptation Fine-Tuning

Diffusion models have achieved remarkable success in generative modeling, yet how to effectively adapt large pretrained models to new tasks remains challenging. We revisit the reconstruction behavior of diffusion models during denoising to unveil the underlying frequency energy mechanism governing this process. Building upon this observation, we propose FeRA, a frequency driven fine tuning framework that aligns parameter updates with the intrinsic frequency energy progression of diffusion. FeRA establishes a comprehensive frequency energy framework for effective diffusion adaptation fine tuning, comprising three synergistic components: (i) a compact frequency energy indicator that characterizes the latent bandwise energy distribution, (ii) a soft frequency router that adaptively fuses multiple frequency specific adapter experts, and (iii) a frequency energy consistency regularization that stabilizes diffusion optimization and ensures coherent adaptation across bands. Routing operates in both training and inference, with inference time routing dynamically determined by the latent frequency energy. It integrates seamlessly with adapter based tuning schemes and generalizes well across diffusion backbones and resolutions. By aligning adaptation with the frequency energy mechanism, FeRA provides a simple, stable, and compatible paradigm for effective and robust diffusion model adaptation.

Bo Yin, Xiaobin Hu, Xingyu Zhou +7

National University of Singapore · University of Electronic Science and Technology of China · Nanyang Technological University +4

Fine-TuningAdaptive Model RoutingDiffusion Model Fine-TuningParameter-Efficient Fine-TuningDiffusion Model Adaptation
AskPDF
Sep 29, 2025

DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space

Existing text-to-image diffusion models excel at generating high-quality images, but face significant efficiency challenges when scaled to high resolutions, like 4K image generation. While previous research accelerates diffusion models in various aspects, it seldom handles the inherent redundancy within the latent space. To bridge this gap, this paper introduces DC-Gen, a general framework that accelerates text-to-image diffusion models by leveraging a deeply compressed latent space. Rather than a costly training-from-scratch approach, DC-Gen uses an efficient post-training pipeline to preserve the quality of the base model. A key challenge in this paradigm is the representation gap between the base model's latent space and a deeply compressed latent space, which can lead to instability during direct fine-tuning. To overcome this, DC-Gen first bridges the representation gap with a lightweight embedding alignment training. Once the latent embeddings are aligned, only a small amount of LoRA fine-tuning is needed to unlock the base model's inherent generation quality. We verify DC-Gen's effectiveness on SANA and FLUX.1-Krea. The resulting DC-Gen-SANA and DC-Gen-FLUX models achieve quality comparable to their base models but with a significant speedup. Specifically, DC-Gen-FLUX reduces the latency of 4K image generation by 53x on the NVIDIA H100 GPU. When combined with NVFP4 SVDQuant, DC-Gen-FLUX generates a 4K image in just 3.5 seconds on a single NVIDIA 5090 GPU, achieving a total latency reduction of 138x compared to the base FLUX.1-Krea model. Code: https://github.com/dc-ai-projects/DC-Gen.

Wenkun He, Yuchao Gu, Junyu Chen +11

Diffusion Model Fine-TuningDiffusion Model Inference AccelerationT2I Diffusion Models
AskPDF
Jun 12, 2025

Anatomy-Grounded Weakly Supervised Prompt Tuning for Chest X-ray Latent Diffusion Models

Latent Diffusion Models have shown remarkable results in text-guided image synthesis in recent years. In the domain of natural (RGB) images, recent works have shown that such models can be adapted to various vision-language downstream tasks with little to no supervision involved. On the contrary, text-to-image Latent Diffusion Models remain relatively underexplored in the field of medical imaging, primarily due to limited data availability (e.g., due to privacy concerns). In this work, focusing on the chest X-ray modality, we first demonstrate that a standard text-conditioned Latent Diffusion Model has not learned to align clinically relevant information in free-text radiology reports with the corresponding areas of the given scan. Then, to alleviate this issue, we propose a fine-tuning framework to improve multi-modal alignment in a pre-trained model such that it can be efficiently repurposed for downstream tasks such as phrase grounding. Our method sets a new state-of-the-art on a standard benchmark dataset (MS-CXR), while also exhibiting robust performance on out-of-distribution data (VinDr-CXR). We further validate our approach through a pilot qualitative study and an experiment on grounded disease classification. Our code will be made publicly available at https://github.com/vios-s.

Konstantinos Vilouras, Ilias Stogiannidis, Junyu Yan +2

School of Engineering, University of Edinburgh · Canon Medical Research Europe Ltd.

Diffusion Model AlignmentPrompt TuningDiffusion Model Fine-TuningMedical Image Grounding
AskPDF
Feb 4, 2025

From Uncertain to Safe: Conformal Adaptation of Diffusion Models for Safe PDE Control

The application of deep learning for partial differential equation (PDE)-constrained control is gaining increasing attention. However, existing methods rarely consider safety requirements crucial in real-world applications. To address this limitation, we propose Safe Diffusion Models for PDE Control (SafeDiffCon), which introduce the uncertainty quantile as model uncertainty quantification to achieve optimal control under safety constraints through both post-training and inference phases. Firstly, our approach post-trains a pre-trained diffusion model to generate control sequences that better satisfy safety constraints while achieving improved control objectives via a reweighted diffusion loss, which incorporates the uncertainty quantile estimated using conformal prediction. Secondly, during inference, the diffusion model dynamically adjusts both its generation process and parameters through iterative guidance and fine-tuning, conditioned on control targets while simultaneously integrating the estimated uncertainty quantile. We evaluate SafeDiffCon on three control tasks: 1D Burgers' equation, 2D incompressible fluid, and controlled nuclear fusion problem. Results demonstrate that SafeDiffCon is the only method that satisfies all safety constraints, whereas other classical and deep learning baselines fail. Furthermore, while adhering to safety constraints, SafeDiffCon achieves the best control performance. The code can be found at https://github.com/AI4Science-WestlakeU/safediffcon.

Peiyan Hu, Xiaowei Qian, Wenhao Deng +8

Work done as an intern at Westlake University · Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing, China · School of Engineering, Westlake University, Hangzhou, China +2

Diffusion Model GuidanceSafety-Critical ControlDiffusion Model Fine-Tuning
AskPDF
Dec 16, 2024

A LoRA is Worth a Thousand Pictures

Low-Rank Adaptation (LoRA) has become a standard approach for parameter-efficient fine-tuning of image diffusion models, with over 200K models trained to replicate artistic styles and shared in online communities. One way to analyze styles of LoRA models is to generate image samples and apply image-based features like CLIP. However, this generation requires additional computation and prompt selection. In this work, we propose analyzing styles directly in the LoRA weight space, bypassing these limitations while achieving better performance even on unseen LoRAs. Since in-the-wild LoRAs lack ground truth model details, we design an experimental setup and show that LoRA weights alone serve as effective feature representations of adapted styles, robust to variations in fine-tuning seeds and image selection. We extensively evaluate using SD1.5 (the most popular base model within online communities) across data domains (fine-art paintings, architectural photos) and diverse prompt settings, with a small-scale SDXL experiment further verifying generalization. These evaluations confirm that our approach, under the selected aspects of variability, preserves fine-grained styles and outperforms image-based methods in clustering and retrieval of unseen LoRAs.

Chenxi Liu, Towaki Takikawa, Alec Jacobson

University of Toronto · Outerport · Adobe Research

Weight-Space AnalysisLow-Rank AdaptationDiffusion Model Fine-Tuning
AskPDF