Diffusion Model Fine-Tuning

Follow this topic Ask about this topic

Momentum

19 papers in the last four weeks, against 2 the four weeks before. 0.2% of all new papers.

Jul 13Week of Sep 28

Latest papers 116

Oct 7, 2026cs.CV

Iris-3B: Going Beyond the Latent with Pixel-Space Diffusion Training, Conversion and Fine-Tuning

Pixel-space diffusion models avoid the lossy VAE of latent models, which suggests an advantage on downstream tasks where fine-grained detail matters. We test this claim along both routes to a pixel-space backbone. We pretrain Iris-3B, a 3B-parameter pixel-space text-to-image transformer, from scratch through a

256\to512\to1024

curriculum, after first ablating the prediction target and representation alignment at

256^2

to decide what to scale. We also convert a pretrained latent model, FLUX.2 Klein base 4B, to pixel space. We fine-tune both families for monocular depth estimation and for image restoration/super-resolution. We find no significant improvement from using a pixel-space generative prior. Fine-tuned for depth with one matched direct-regression recipe, Iris-3B is level with the latent FLUX.2 Klein and the converted pixel FLUX.2 Klein falls behind it, and on

4\times

DIV2K restoration neither pixel model beats a latent FLUX.2 Klein fine-tune, the converted one trailing it slightly. We document the recipes, the failure modes and the remaining confounds behind this negative result. Nevertheless, Iris-3B shows that pixel-space pretraining with the pixel-transformer (PiT) head of PixelDiT scales to 3B parameters and to text-to-image quality competitive with latent models, matching Qwen-Image on OneIG under the official evaluators at

1024^2

. We release its weights and training code in the hope that they help pave the way for further work on pixel-space generation.

Hanqiu Li Cai, Chema Garabito

SperidLabs

T2I Diffusion Models Pixel-Space Diffusion Models Diffusion Transformer Diffusion Model Fine-Tuning Diffusion Models for Image Restoration

Oct 6, 2026cs.LG

Are Parameter-Efficient Fine-tuning Methods Really Different?

Parameter-efficient fine-tuning (PEFT) offers many parameterizations, yet their methodological and functional differences remain unclear. We compare six methods in language and diffusion models to examine how their parameterizations relate to task performance, forgetting, and changes in pretrained weight geometry. Motivated by the spectrum-preserving design of orthogonal fine-tuning (OFT), we first ask whether spectral preservation is itself important for adaptation and retention. We find that the selected LoRA-family methods also approximately preserve pretrained geometry, and that restoring their slightly drifted singular-value spectra largely preserves task performance, questioning the necessity of explicit geometric preservation. Beyond this, we observe that some methods exhibit distinct adaptation--retention trade-offs that vary across settings: LoRA most consistently limits forgetting at competitive performance, DoRA achieves higher mean task scores than LoRA in most comparisons, while PiSSA often incurs greater retention costs. Further intervention experiments suggest that while performance gains from different PEFT methods can be attributed to modifications in different groups of spectral components, we consistently find that restoring dominant rather than intermediate or trailing components produces the largest mean reduction in general-text NLL or base-image drift. Together, these results motivate evaluating geometric constraints through their functional consequences rather than preservation alone. Code is available at https://github.com/Kuaaannn/PEFT_methods.

Yikuan Li, Pinyan Lu, Fanghui Liu

Department of Statistics, The University of Warwick, UK. · Key Laboratory of Interdisciplinary Research of Computation and Economics, Shanghai University of Finance and Economics, China. · School of Mathematical Sciences, Institute of Natural Sciences and MOE-LSC, Shanghai Jiao Tong University, China.

Parameter-Efficient Fine-Tuning Low-Rank Adaptation Catastrophic Forgetting Diffusion Model Fine-Tuning

Oct 6, 2026cs.LG

MaRK: Markov-adapted Recurrent Kernels for Dynamic Operator Conditioning in State Space Models

State Space Models (SSMs) offer an efficient alternative to Transformers for sequence modeling, yet conditioning pre-trained SSMs for iterative generation typically operates outside the recurrent operator, through input injection or activation modulation. While such mechanisms expose the model to conditioning information, they leave the underlying temporal dynamics fixed. We introduce MaRK (Markov-adapted Recurrent Kernels), a dynamic operator-conditioning framework that maps context vectors directly into bounded modulations of a frozen SSM's recurrence (

A

), read-in (

B

), read-out (

C

), skip (

D

), and discretization (

Δ

) parameters. Viewed through the lens of LPV-SSM systems, MaRK induces a context-indexed family of Markov parameter sequences, allowing each diffusion timestep to reshape the model's input-output memory kernel. We instantiate MaRK on a frozen 111M-parameter Hydra SSM backbone and study three adapter geometries: Hypernet, Chebyshev polynomial, and Discrete Cosine Transform kernels. Since these adapters modify the Markov parameter sequence through low-rank auxiliary maps on the frozen backbone, parameter-efficient fine-tuning arises as a structural consequence of the adaptation mechanism itself, requiring only 6.3--11M trainable auxiliary parameters to transition from a bidirectional objective to an iterative diffusion regime. The bounded recurrence parameterization further yields an analytic Affine Quadratic Stability certificate for the modulated recurrence. Through synthetic LPV recovery experiments and Markov-operator diagnostics, we show that MaRK recovers coordinate-invariant temporal operators under matched assumptions and produces distinct, stable timestep-conditioned memory profiles. Empirically, the Chebyshev variant yields the strongest performance, achieving an average validation loss of 2.55, followed by the DCT (2.59) and Hypernet (3.77) geometries.

Syed Ibrahim Omer, Ginny Y. Wong, Xiangyu Zhao

Department of Data Science City University of Hong Kong

State Space Models Diffusion Model Fine-Tuning

Oct 6, 2026cs.CV

LiDAR Resolution Recovery via Foundation-Model-Guided Diffusion

High-beam-count LiDAR sensors are costly, yet many perception pipelines require dense angular sampling. Using a pretrained Stable Diffusion model as the backbone, we fine-tune a LiDAR-conditioned depth model with pseudo-depth targets from a 2D foundation model. During training, the LiDAR conditioning is randomly decimated at different beam budgets. We then investigate how much of a LiDAR scan can be recovered from heavily decimated input and characterize performance across the input beam budget. We evaluate against physically held-out real beams on nuScenes and report recovery separately from fit accuracy. Our model yields its largest advantage in very sparse regimes, achieving a

δ_{1.25}

accuracy of

66.8

% from

4

-beam input where scattered interpolation reaches only

45.1

%. A class-stratified error breakdown further reveals that planar surfaces recover first while objects introducing depth discontinuities degrade earliest. Together, these results quantify the recovery/resolution trade-off for foundation-model-guided LiDAR enhancement.

Samed Doğan, Nico Leuze, Alfred Schöttl

Dept. of Electrical Engineering and Information Technology Munich University of Applied Sciences, 80335 Munich, Germany

Diffusion Model Guidance Depth Completion Diffusion Model Fine-Tuning

Oct 5, 2026cs.SD

Adapting a Latent Audio Diffusion Model to Historical Guqin Recordings: A Listening-Driven Case Study

We report a small-data case study in adapting a pretrained latent audio diffusion model to the guqin, the seven-string Chinese zither, aiming at an "AI radio" that plays guqin-style music without end. From a library of historical recordings we curate 412 solo performances (42.5 h, 61 performers) and split them by composition. We fine-tune a rank-16 DoRA adapter on Stable Audio 3 Medium using its continuous latents, masking weighted towards continuation, captions that combine researched notes on each piece with mood tags and an automatically estimated pentatonic mode, and random-length crops, stopping when held-out loss stops improving. Seven blind listening studies by one expert listener guided every decision. The final adapter was rated highest for continuing unseen pieces (3.9/5, against 3.4 for the best earlier adapter) and 4.6/5 for generating from free-written scene descriptions. Negative results are equally informative: a from-scratch autoregressive model over the same latents produced no recognisable timbre, a signal-level friction-noise measure correlated with the listener's complaints in the wrong direction, and a pentatonic-fit measure tracked ratings overall but barely within a group of candidates. Chaining continuations for long playback exposed a silent tail on every generated clip and a loudness feedback loop, both with simple fixes. With one listener and at most ten clips per condition, no paired difference is statistically significant; we present an exploratory record of what helped, what did not, and why.

Huanchen Cai

Division of Speech, Music and Hearing (TMH) KTH Royal Institute of Technology, Stockholm, Sweden

Text-to-Music Generation Music Generation Evaluation Audio Diffusion Models Diffusion Model Fine-Tuning Music Generation

Oct 1, 2026cs.CV

CLASP: Continual Low-rank Adapters for Spatially Placed Concepts from One Hypernetwork

Continual personalization of text-to-image diffusion models requires sequentially acquiring new concepts while retaining previously learned ones. However, existing methods either suffer from catastrophic forgetting or rely on storing additional concept-specific parameters and spatial components, causing their parameter footprint to grow with the concept stream. This limits their ability to scale to long sequences of personalization tasks. We propose a rehearsal-free approach that uses a single fixed-size hypernetwork to continually personalize a frozen diffusion model. Instead of expanding the model as new concepts are acquired, the hypernetwork dynamically produces the concept-specific adaptations required for personalization while preserving previously learned concepts. Our framework further integrates spatial control into the personalization process, allowing users to specify where a personalized concept should appear without introducing additional per-concept components. This formulation enables continual personalization with a parameter footprint that remains independent of the number of learned concepts, aside from compact concept representations. Experiments demonstrate strong retention of previously learned concepts and reliable spatial grounding, matching or improving upon existing methods while scaling effectively to long streams of personalization tasks.

Wojciech Gromski, Patryk Krukowski, Jan Miksa +2

Wrocław University of Science and Technology · IDEAS Research Institute · Jagiellonian University, Krak´ow +2

Subject-Driven Image Generation T2I Generation Diffusion Model Fine-Tuning Low-Rank Adaptation Hypernetworks Diffusion Models

Sep 30, 2026cs.LG

Fenchel Tilting: Weighted Correction for Efficient Finetuning of Generative Models

Adapting a pretrained generative model to an arbitrary preference expressed as a utility function underlies reward alignment, guided design, and constraint satisfaction, enabling diverse applications. Existing fine-tuning methods trade off generality against computational cost: they either restrict the family class of supported preferences to keep optimization simple or preserve generality at the expense of efficiency. We introduce Fenchel Tilt Flow Control (FTFC), which decouples utility optimization from generative-model fitting. FTFC first optimizes for a target distribution by jointly fitting an effective reward and density-ratio weights on pretrained samples. Method combines the utility's variational structure with Fenchel duality, supporting general

f

-divergence penalties that determine how rewards are transformed into an distribution-correction weights. These weights are then frozen and used to modify a diffusion or flow model in a single stage of importance-weighted denoising or flow matching, without differentiating through sampling trajectories. We establish exact duality for concave utilities under suitable conditions and show that weighted fitting reproduces the optimal target distribution for a given utility. Across image and molecule generation benchmarks, FTFC improves over baselines on diverse preference functions, while also being up to

20\times

more efficient. roposed method enables adaptation beyond expected-reward maximization without complex optimization, while preserving robustness for more general class of the utility functions compared to baselines.

Maksim Bobrin, Maksim Zhdanov, Dmitry Dylov

AXXX · Applied AI Institute · MBZUAI

Diffusion Model Alignment Constrained Generative Modeling Fine-Tuning Diffusion Model Fine-Tuning

Sep 30, 2026cs.CV

CAST: Causal Advantage-Structured Training with Spatially Grounded Compositional Rewards for Diffusion Models

Online reinforcement learning has been extended to flow matching for diffusion model (DM) image generation. However, this paradigm faces three limitations: (1) Window selection. Existing methods manually set the stochastic differential equation (SDE) sampling window, i.e., the denoising steps where exploration noise is injected. We instead determine it from each model's denoising trajectory. (2) Reward saturation. Current methods rely on scoring models trained on human annotations; we find that such scores are extremely high and nearly indistinguishable on the latest SOTA open-source DMs, making advantage estimation largely ineffective. (3) Sample inefficiency. A single scalar reward collapses different failure modes into almost identical scores, leaving minimal gradient guidance for targeted improvement. To address these issues, we propose CAST (Causal Advantage-Structured Training), an RL fine-tuning method for pretrained DMs, which (1) identifies the denoising step at which each model fixes the objects and their spatial arrangement in the image and uses that timing to set the SDE window, (2) decomposes each prompt via Causal Scene Graphs (CSG) into verifiable-atoms, i.e., minimal semantic units such as an object, count, attribute, or spatial relation that can each be checked independently, and rewards each atom separately, and (3) projects the signed atom-level advantages into pixel space through teacher-forced attention and uses them to spatially weight the SDE policy objective. We fine-tune two of the strongest open-source DMs, FLUX.2-dev and Qwen-Image-2512, with CAST, and evaluate them on GenEval 2, a compositional benchmark, and on Qwen-Image-Bench for overall quality. Within almost the same training budget, CAST's improvement over the base model on the most challenging GenEval 2 prompts is up to 3.07x that of Flow-GRPO, while overall generation quality also improves.

Shu Yu, Chaochao Lu

Shanghai Artificial Intelligence Laboratory, Shanghai, China · Shanghai Innovation Institute, Shanghai, China · Fudan University, Shanghai, China

Diffusion Model Fine-Tuning Compositional T2I Generation RL for Diffusion Models Flow-Based Generative Modeling Reward Design for RL

Sep 30, 2026cs.CV

ReGain: Restoring Subject Fidelity in Personalization on Synthetic Images

Text-to-image diffusion models are personalized to a subject by DreamBooth fine-tuning on a handful of its images. Increasingly, these images come from a diffusion model rather than a camera. We show that fine-tuning on such synthetic images degrades subject fidelity, producing oversaturated color and excess high-frequency detail. To isolate the cause, we fine-tune two models from the same base model with the same DreamBooth recipe, one on real photos of a subject and one on synthetic images of that subject generated by the first. We trace the degradation to classifier-free guidance (CFG). For the model personalized on synthetic images, the angle between the conditional and unconditional noise predictions, and with it the norm of their difference, is much larger than for the model personalized on real photos. This inflation grows toward high frequencies and also appears at other prompts semantically close to the subject, such as its class noun, but not at unrelated ones. We propose ReGain, a training-free correction applied at sampling time that measures how much each frequency band of the guidance is inflated relative to the base model and scales that band down accordingly. ReGain needs no real photos. On Stable Diffusion v1.5, ReGain closes 51-64% of the subject-fidelity gap to the model personalized on real photos, as measured by DINO, DINOv2 and CLIP-I. It also improves subject fidelity on SDXL and SD 3.5 and preserves text alignment on all three backbones.

Shubhang Bhatnagar, Ishan Bhatnagar, Viraj Shah +1

University of Illinois Urbana-Champaign

Diffusion Model Guidance Classifier-Free Guidance Subject-Driven Image Generation Diffusion Model Fine-Tuning

Sep 29, 2026cs.AI

Fine-Tuning Diffusion Language Models with Context Selection and Target Weighting

Supervised fine-tuning of discrete diffusion language models masks some response tokens and trains the model to recover their original values from the visible context. The masking pattern therefore determines both the context available to the model and the tokens it learns to predict. Uniform random masking does not explicitly account for the interaction between these choices. We introduce GoldiMask, which selects tokens to reveal as context by approximately maximizing a submodular objective. This objective uses model signals to balance the benefit of revealing tokens against their value as prediction targets. GoldiMask then weights the remaining targets according to how they benefit from the selected context and their remaining learning potential. Across three backbones and three training datasets, GoldiMask achieves the highest average accuracy in most evaluated settings, demonstrating gains on both reasoning and code generation. Component ablations show that both context selection and target weighting contribute to the gains. GoldiMask also reduces decoding iterations on GSM8K and MATH-500 under confidence-threshold parallel decoding, while maintaining comparable accuracy at higher confidence thresholds.

Loay Mualem, Lluís Pastor-Pérez, Vinh Tong +4

IMPRS-IS · Institute for AI, University of Stuttgart · Bitdefender

Masked Diffusion Models Diffusion Model Fine-Tuning Diffusion Language Models

Sep 29, 2026cs.CV

ExploreNet: Learning Where to Explore in Diffusion GRPO

Group-relative RL methods such as Flow-GRPO post-train image generators by exploring with isotropic Gaussian noise added at every denoising step. This noise decides which rollouts the model learns from, yet it perturbs every channel and spatial position of the latent equally. In this paper, we instead show that latent elements differ in how much they change the generated image, so exploration should adapt to these differences. We introduce EXPLORENET to learn an adaptive exploration distribution. EXPLORENET is a policy that predicts a noise scale for every latent element from the current latent, the denoising step, and the prompt, before any reward is observed; it is trained on the reward spread of each rollout group and discarded after training, leaving inference unchanged. On Stable Diffusion 3.5 Medium, EXPLORENET improves held-out GenEval2 by 14% over Flow-GRPO, transfers to two independent compositional benchmarks and five preference and image-quality models, and reaches a 67.2% human preference win-rate. Overall, across our group-relative diffusion RL experiments, we find that exploration is learnable, the shape of the exploration distribution outweighs its magnitude, and rollout quality is more effective than rollout quantity.

Shuyue Stella Li, Xiaochuang Han, Yulia Tsvetkov +1

University of Washington

Diffusion Model Fine-Tuning RL Exploration RL for Image Generation RL for Diffusion Models T2I Diffusion Models

Sep 29, 2026cs.CV

Adversarial Training for Pixel Diffusion

Pixel diffusion models generate RGB images directly, avoiding the bottleneck of an autoencoder, yet their outputs still systematically underrepresent fine-scale natural-image statistics. We show that adversarial learning provides an effective post-training correction for this deficiency. Starting from a pretrained model, we retain its original diffusion or flow-matching objective and add an adversarial loss to the predicted output at non-high-noise timesteps, leaving the model architecture and sampling procedure unchanged. To our knowledge, this is the first systematic study of adversarial post-training for pixel diffusion. Across two pixel backbones, the method jointly improves distribution fidelity, coverage, prompt alignment, and perceptual quality. We further investigate why it works. Frequency-band and power-law analyses show that the original models systematically underproduce natural-image high-frequency content, while adversarial post-training restores this missing spectral power. In contrast, perceptual loss also increases high-frequency content but sacrifices distribution fidelity and prompt alignment. Nearest-neighbor, recall, and matched no-GAN SFT controls further rule out memorization, mode dropping, and additional optimization as simple explanations. Finally, we examine the boundary of this effect. Under the tested latent diffusion configurations, the same procedure does not produce comparable joint gains and adds almost no decoded high-frequency power. These results identify direct output access to the image statistics being corrected as a key factor governing when adversarial post-training succeeds.

Xin Lin, Zhifei Zhang, Yuqian Zhou +4

UC San Diego · Adobe Research · Work done during an internship at Adobe Research +1

Adversarial Training Pixel-Space Diffusion Models Diffusion Model Fine-Tuning

Sep 29, 2026cs.LG

On Trajectory-Aware Training for Masked Diffusion Language Models

Masked diffusion models (MDMs) generate text by unmasking several tokens per step, but they are trained and sampled under different conditions. The model is trained on randomly masked sequences, whereas inference follows a trajectory shaped by the model's own predictions. Additionally, each step has no access to what the previous one computed. Recent methods narrow these limitations from separate angles, leaving open how these choices interact. We introduce PUMBA, a unified framework for trajectory-aware training that trains the denoiser on consecutive steps of policy-induced trajectories, passes information between steps, and optimizes them jointly by backpropagation through time. A controlled study of this design space shows that i) exact train--inference alignment fails due to local overfitting, whereas a looser alignment still brings training masks closer to those seen at inference; ii) passing continuous information outperforms discrete gradient estimators through the commitment at each step; and iii) performance improves as backpropagation through time spans more steps, which we support theoretically. Combined, these components match the best checkpoint of a same-size autoregressive model. Building on these findings, we scale PUMBA to supervised fine-tuning of LLaDA-8B, where it improves the trade-off between performance and number of function evaluations (NFEs) in both full-canvas and block diffusion generation. At matched performance, it needs up to 22% fewer NFEs than standard fine-tuning with twice the budget in full-canvas generation, and up to 26% fewer than standard fine-tuning for the same number of steps in block diffusion.

Manuel Madeira, Amitis Shidani, Alice Bizeul +6

EPFL · Apple · Georgia Institute of Technology

Masked Diffusion Models Diffusion Language Model Inference Diffusion Model Fine-Tuning Diffusion Language Models

Sep 29, 2026cs.AI

MeanFlowAdvantage: Stable Reward Fine-Tuning for Few-Step Average-Velocity Generators

MeanFlow enables efficient few-step generation by predicting interval-average velocities, but this representation creates a mismatch for reward fine-tuning: existing advantage-based objectives are typically defined on instantaneous velocities or equivalent

x_0

-space predictions, whereas inference directly uses the learned average-velocity map. We introduce MeanFlowAdvantage, a signed advantage-weighted least-squares objective for average-velocity generators. Our key construction uses a shared, detached MeanFlow derivative correction to express the reward objective in prediction space while making rollout and reference regularization exact penalties on the average-velocity network deployed at inference. The resulting formulation preserves MeanFlow's native few-step sampler and provides a direct mechanism for transferring reward improvements to the deployed flow map. On SD3.5-Medium, MeanFlowAdvantage improves all eight reported metrics over the matched four-step MeanFlowNFT baseline and, with only four NFEs, matches or exceeds the 40-step DiffusionNFT baseline on six of eight metrics. The same objective also transfers to DNA promoter design, where it supports both teacher-free on-policy RL for a generator defined on a manifold and teacher-guided reward-graded distillation, with the latter yielding the lowest one-step Sei profile MSE among the compared configurations.

Haocheng Tang, Tianchi Xie, Xingqiao Lin

Northeastern University Boston, MA 02115, USA · Tsinghua University Beijing, 100084, PRC · Carnegie Mellon University Pittsburgh, PA 15213, USA

Flow Matching MeanFlow Few-Step Diffusion Sampling Diffusion Model Fine-Tuning Flow Map Learning RL for Diffusion Models+1

Sep 28, 2026cs.CV

Triangular Resampling for Long-Horizon Motion Generation

We introduce Triangular Resampling (TR), a post-training method for mitigating long-horizon error accumulation in motion diffusion models. Built on FloodDiffusion's triangular denoising schedule, TR addresses the mismatch between ground-truth-derived training windows and model-generated inference states. Replacing only completed motion history leaves this mismatch unresolved in partially denoised states within the active window. TR therefore extends rollout-based training to these states, using ground-truth clamping to limit excessive drift. For each replayed sample, TR draws one denoising threshold, shared across latent positions and replay updates, and replays multi-step triangular denoising without gradient tracking. After each update, states below the threshold are replaced with noise-matched ground truth, while those at or above it retain model predictions. The resulting latent window enters the standard training update. This rollout construction supports both supervised training (TR) and distribution matching (TR-DMD). On 120-second motion generation from HumanML3D test prompts, TR and TR-DMD achieve state-of-the-art FID AUC within their respective non-DMD and DMD comparison groups. Supervised TR reduces FID AUC by 40.9% and FID degradation slope by 55.3% relative to matched post-training without replay.

Kunhang Li, Yiyi Cai, Xiangyue Zhang +5

The University of Tokyo · Alaya Lab · Japan Advanced Institute of Science and Technology

Motion Diffusion Models Human Motion Generation Diffusion Model Fine-Tuning Diffusion Models for Motion Generation Diffusion Models

Sep 27, 2026cs.CV

Revisiting Diffusion Fine-Tuning for Unsupervised Domain Adaptation

Diffusion-based unsupervised domain adaptation (UDA) improves cross-domain transfer by generating target-specific synthetic data for downstream adaptation. Existing methods are largely designed for single-target adaptation: when a model trained on one labeled source domain must be adapted to multiple unlabeled target domains, they typically require separate diffusion fine-tuning for each source--target pair, causing training, storage, and deployment costs to grow with the number of targets. In this paper, we study multi-target data generation for diffusion-based UDA, where a single source-guided diffusion fine-tuning process is reused to generate target-specific synthetic data for multiple target domains. We propose MUSE (Multi-target UDA-oriented Synthesis with Efficient diffusion fine-tuning), a decoupled adaptation framework that separates source-supervised semantic adaptation from target-specific style adaptation. MUSE uses a shared semantic branch updated by labeled source data and target-private style branches specialized to individual target domains, enabling target-specific generation while avoiding repeated source-guided fine-tuning for each target. Experiments on standard UDA benchmarks show that MUSE achieves a stronger accuracy--efficiency trade-off than repeated per-target diffusion adaptation, reducing diffusion fine-tuning cost while improving average target-domain accuracy. The project page is available at https://xuanqi99.github.io/MUSE/.

Xuan Qi, Yi Wei, Daniele Berardini +2

AI for Good (AIGO), Istituto Italiano di Tecnologia, Genoa, Italy · DITEN, University of Genoa, Genoa, Italy · National Key Laboratory of Novel Software Technology, Nanjing University, China +3

Unsupervised Domain Adaptation Domain Adaptation Diffusion Model Fine-Tuning

Sep 27, 2026cs.LG

Elucidating the Design Space of Regression-based Diffusion Reinforcement Learning

A nascent family of methods that forgoes the policy gradient and reweights a supervised regression instead has garnered momentum in reinforcement learning for diffusion and flow models. DiffusionNFT, FlowAWR, and RAM are representative regimes with contrasting motivations. It is yet opaque what, if anything, they share. We substantiate that each is the solution of one divergence-constrained reward-maximization problem, and they are differentiated only by the convex generator that defines the constraint. Under the unified modeling framework, we unravel the relaxations that prior art made during building the advantage-embedded regression target: approximating the KKT condition and posterior normalizer for the linear and exponential tilt shapes DiffusionNFT and FlowAWR respectively, while preserving the exact sparsemax projection onto the probability simplex for linear tilt leads to another superior model type in this work. Beyond the theoretical underpinnings, we further empirically investigate the design space and shed light on the training recipe for regression-style diffusion RL. Retaining the merits discovered during our exploration gives rise to DiffusionRFT, our paradigm that converges faster, trains more stably, and attains the top performance.

Toyota Li, David Zhao, Alan Zhao

Tencent

Reinforcement Learning Constrained RL Diffusion Model Fine-Tuning RL for Diffusion Models

Sep 25, 2026cs.LG

DOHF: Online Diffusion Fine-tuning with Doob's $h$ -transform Guidance

Reward-based diffusion fine-tuning faces practical challenges when desirable outcomes are rare or conditioning corrections are costly to estimate. In this work, we propose Diffusion Online

h

-guidance Fine-tuning (DOHF), which turns Doob's

h

-transform into a practical online training algorithm. DOHF assigns optimality weights to generated samples, estimates the normalized local correction

\nabla\log h

under the current rollout policy, and distills it directly into the generative model. Theoretically, we characterize the population-optimal DiffusionNFT update as well as the various classfier free guidance methods through a unified

h

-transform perspective. Methodologically, our framework accommodates black-box and non-differentiable rewards without additional network evaluations. We further show improved alignments under three empirical scenarios. Our work demonstrates how adapting probabilistic conditioning through inexpensive estimation and iterative distillation can improve generative learning across statistical sampling and visual generation.

Zhengyi Guo, Jiayuan Sheng, Wenpin Tang +1

Diffusion Model Alignment Diffusion Model Guidance Doob H-Transform Diffusion Model Fine-Tuning RL for Diffusion Models

Sep 21, 2026cs.LG

Corrective Forcing: Unified Post-Training for Diffusions and Flows in Generative Speech Enhancement

Diffusion and flow models, as promising generative paradigms for speech enhancement, face a training--inference mismatch: training uses analytical path states, whereas inference recursively evaluates models on self-generated rollout states along discretized sampling trajectories. This mismatch causes prediction and discretization errors to accumulate. To address it, we introduce Corrective Forcing (CoF), a post-training paradigm that forces diffusion and flow models to learn from self-generated rollouts and correct their predictions. CoF corrects clean-speech predictions on rollout states toward the ground truth under dynamic sampling schedules, exposing the model to varying inference conditions. It further regularizes local evolution using locally corrected counterfactual transitions as references for factual transitions. By expressing model outputs through a shared clean-speech prediction parameterization, CoF applies the same post-training objective across diffusion and flow formulations. Experiments with SB-VE and OT-CFM demonstrate improvements in perceptual quality and reconstruction fidelity, together with robust performance across different numbers of sampling steps.

Qing Yao, Lijian Gao, Qirong Mao

School of Computer Science and Communication Engineering, Jiangsu University Jiangsu Engineering Research Center of Big Data Ubiquitous Perception and Intelligent Agriculture Applications Provincial Key Laboratory of Computational Intelligence and New Technologies in Low-Altitude Digital Agriculture Zhenjiang, China

Diffusion Model Fine-Tuning Speech Enhancement Flow-Based Generative Modeling

Sep 14, 2026cs.LG

Principal-timestep Restricted Init via Sparse Matrix-decomposition in Flow-matching

Flow-matching diffusion models have recently emerged as a strong paradigm for high-fidelity visual generation. However, their prohibitively high fine-tuning cost limits scalability to downstream tasks. While Low-Rank Adaptation (LoRA) combined with spectral initialization has demonstrated accelerated convergence and improved performance in autoregressive language models by better aligning gradient directions, we find that it fails to deliver similar gains in diffusion fine-tuning, often yielding marginal or even negative improvements over vanilla LoRA.We attribute this discrepancy to a fundamental mismatch between LoRA's low-rank parameterization and the intrinsically high-rank gradients induced by the flow-matching objective. In particular, stochastic timestep sampling introduces directionally heterogeneous gradient signals across training steps, leading to misaligned updates under low-rank constraints.To address this issue, we propose Prism-LoRA,a Principal-timestep Restricted Init via Sparse Matrix-decomposition framework that improves gradient alignment during fine-tuning. Our method consists of two key components: (i) principal timestep selection, which restricts initialization gradients to a subset of dominant timesteps to suppress effective gradient rank, and (ii) principal channel filtering, which removes task-irrelevant channels, enabling the one-step spectral initialization gradient to better align with the long-horizon optimization trajectory. Extensive experiments demonstrate that our method consistently improves both convergence speed and final performance across multiple diffusion fine-tuning benchmarks, including subject-driven generation, controllable generation, and deblurring, achieving not only performance improvement but also earlier stages of convergence over baseline LoRA and other spectral-init methods.

Jiayang Gu, Zheng Fang, Lichaun Xiang +3

University of Warwick · Bytedance

Flow Matching Diffusion Model Fine-Tuning Low-Rank Adaptation Spectral Methods

Sep 14, 2026cs.CV

DNF-SR: Dual-Input and Negative-Aware Feature Fine-Tuning for Real-World Image Super-Resolution

Benefiting from the powerful generative priors of diffusion models, diffusion-based real-world image super-resolution (Real-ISR) methods have demonstrated impressive performance.To achieve efficient Real-ISR, several recent works have designed one-step diffusion-based models.Howerver, unmediatedly feeding LR into a diffusion model creates a distributional gap with the model's original input.A straightforward approach to reduce the distribution gap is to introduce noise to the LR latents. However, directly adding noise inevitably corrupts the content of the LR images.In this study, we propose DNF-SR, a Dual-input and Negative-aware Feature fine-tuning method for Real-ISR.Specifically, we use a dual-input strategy that concatenates the original LR image with the noisy LR input and feeds them into a diffusion-based image editing model, ensuring both high-fidelity one-step super-resolution and improved perceptual and content consistency.Additionally, the noise present in the noisy LR input introduces randomness and diversity into the outputs. We exploit this property and propose a post-training optimization method, Negative-aware Feature Fine-Tuning (NF2T), which guides the model toward producing higher-quality results.NF^2T classifies multiple outputs into positive and negative subsets and then defines implicit policy improvement directions in both the image and feature spaces, thereby further enhancing the stability of the optimization.Extensive experiments show that DNF-SR outperforms other methods.Code will be released.

Shuhao Han, Wenjie Liao, Hayden Vance +4

VCIP, CS, Nankai University · Independent Researcher

Diffusion Models for Image Restoration Diffusion Model Fine-Tuning Image Super-Resolution

Sep 14, 2026cs.CV

GRACE: Adaptive Concept Erasure with Geometry-Guided Retention in Diffusion Models

Text-to-image (T2I) diffusion models inevitably internalize sensitive or non-compliant concepts from large-scale pretraining data, necessitating post-hoc concept erasure. However, existing erasure methods often lack explicit constraints on parameter updates, leading to over-intervention and unintended semantic drift. In addition, many methods rely on manually crafted counterfactual supervision, such as surrogate prompts, which incurs substantial data construction costs that limit scalability to new concepts. To address these limitations, we propose GRACE, a structured concept erasure framework designed to enable localized and selective intervention. Specifically, we introduce a semantically weighted sensitive subspace estimation to precisely lock intervention directions, and employ lightweight subspace-constrained adapters to prevent global semantic disturbance. To eliminate the dependency on manual prompt engineering, we design an automatically decoupled safe-anchor mechanism. To mitigate semantic drift induced by excessive intervention, we introduce an energy-driven dynamic gating mechanism that adaptively controls the timing and strength of intervention at inference. Extensive experiments demonstrate that our method achieves a superior balance between erasure effectiveness and generation fidelity. Compared with the average performance of five state-of-the-art (SOTA) concept erasure methods, our method improves the fine-grained NSFW reduction rate by

17.86\%

, while reducing the macro-averaged target CLIP Score and preservation-oriented Fr'echet Inception Distance (FID) by

4.75\%

and

50.58\%

, respectively, indicating stronger concept suppression with substantially improved preservation of the original model's generative utility.

Qinghui Gong, Yihuai Liang, Yuanlun Xie +3

School of Information Science and Technology, Southwest Jiaotong University, Chengdu 611756, China · School of Electronic Information and Electrical Engineering, Chengdu University, Chengdu, China · Key Laboratory of Intelligent Control and Optimization for Industrial Equipment of Ministry of Education, Dalian University of Technology, Dalian, China +1

Diffusion Model Fine-Tuning Diffusion Model Unlearning Concept Erasure T2I Diffusion Models

Sep 12, 2026cs.AI

Understanding LoRA Rank Trade-offs in Diffusion Model Fine-Tuning

Selecting LoRA rank for diffusion fine-tuning requires balancing quality and compute cost. We present a controlled study on CIFAR-10 using a DDPM U-Net with ranks {2,4,8,16,32}, fixed optimization settings, and a reproducible local-folder pytorch-fid protocol. We report FID, trainable parameters, runtime, and GPU memory, then validate trends with extended-budget DDPM runs (20 epochs; ranks 4/8/16) and a Tiny DiT backbone (10 epochs; ranks 4/8/16). Results show moderate ranks are most efficient: rank 4 achieves the best DDPM FID (124.1380), rank 8 is close (124.2136), and higher ranks provide limited gains despite larger adaptation cost. These findings support small-to-moderate ranks as practical defaults under fixed training budgets.

Iman Khazrak, Narges Nejad, Mostafa M. Rezaee +1

Department of Computer Science, Bowling Green State University, Bowling Green, OH, USA. · Department of Management and Marketing, Angelo State University, San Angelo, TX, USA.

Diffusion Model Fine-Tuning Low-Rank Adaptation

Sep 8, 2026cs.CV

Marigold V2: Revisiting Diffusion Transformers for Monocular Depth Estimation

Monocular depth estimation is a ubiquitous yet highly ill-posed computer vision task, with downstream applications in scene reconstruction, computational photography, and robotics, among others. Despite the field's maturity, recent models still struggle to generalize to out-of-distribution inputs and to produce sharp and detailed depth maps. In this paper, we revisit Marigold, a set of techniques for repurposing modern image generation and editing models, powered by the diffusion transformer (DiT) architecture, into state-of-the-art monocular depth estimators. Our recipes target single-step inference from pretrained multi-step flow-matching models, with quantization where needed, preserving model capacity while remaining cheap to run. We analyze the artifacts of naive training and identify two effective remedies: aligning the model's internal representations with semantic features extracted from ground-truth, and adopting a 2-stage fine-tuning protocol built around a novel Sinkhorn-based loss. The results are crisper, cleaner depth maps that generalize well out-of-distribution, with 16-26% improvement in AbsRel over the previous best on KITTI and ETH3D. Qualitatively, our model resolves fur, foliage, and hair-thin edges that have eluded prior models. Furthermore, Marigold V2 achieves state-of-the-art results when applied to other dense regression tasks, such as surface normals estimation and intrinsic image decomposition. Project website: https://hf.co/spaces/huawei-bayerlab/marigold-v2-web

Igor Pavlovic, Thiemo Wandel, Anton Obukhov +6

EPFL, HUAWEI Bayer Lab, Switzerland · HUAWEI Bayer Lab, Switzerland · University of Bologna, Italy +1

Diffusion Transformer OOD Generalization Monocular Depth Estimation Surface Normal Estimation Diffusion Model Fine-Tuning

Sep 1, 2026cs.CV

Gaussian Core LoRA: Distribution-Aware Dynamic Adaptation for Broad Concept Erasure

Concept erasure aims to suppress unsafe, privacy-sensitive, or undesirable generations in text-to-image diffusion models while preserving benign semantics, visual quality, and deployment efficiency. Existing adapter-based methods, such as Low-Rank Adaptation (LoRA), typically freeze the diffusion backbone and learn lightweight parameter updates to steer generation away from target semantics. However, these methods usually assign a static semantic erasure direction to each target concept. This assumption is overly coarse for broad and complex target concepts, since a concept often contains multiple latent semantic prototypes involving different objects, scenes, or relations, and requires different local erasure directions. A single LoRA update averages these heterogeneous erasure demands, leading to under-erasure on difficult prototypes and over-editing of nearby benign semantics. To address this limitation, we propose Gaussian Core LoRA, a distribution-aware low-rank adaptation framework. It fits a Gaussian mixture model in the prompt feature space to estimate latent semantic prototypes within the target concept. During inference, each input prompt is projected into this feature space to compute its Gaussian posterior responsibilities, which condition the core generator to produce a prompt-specific, norm-bounded residual reconfiguration of the shared LoRA rank space. This enables prototype-adaptive erasure with a single lightweight adapter. Compared with the strongest baseline on each metric, Gaussian Core LoRA reduces average Attack Success Rate (ASR) by 7.95%, lowers COCO Fr'echet Inception Distance (FID) by 14.72%, and improves CLIP Score by 4.98%. Further experiments show robustness to adversarial prompts, scalability to multi-identity and multi-style erasure, and compatibility with SDXL and FLUX.

Qinghui Gong, Xunlei Chen, Yu-Xuan Zhang +2

Southwest Jiaotong University · University of Electronic Science and Technology of China

Diffusion Model Fine-Tuning Low-Rank Adaptation Diffusion Model Unlearning T2I Diffusion Models

Aug 11, 2026cs.CV

AdvFD: Boosting Visual Generation via Adversarial Fr'echet Distance Loss

Fréchet distance has recently emerged as an effective distribution-level objective for generator post-training, complementing the conventional sample-level diffusion and flow-matching losses. However, directly optimizing Fréchet objectives can cause Fréchet hacking. The target metrics keep improving, but visual quality and Fréchet alignment in other feature spaces may stagnate or deteriorate. We attribute this failure to the static pretrained feature spaces used by existing Fréchet losses. These feature spaces provide incomplete and fixed views of the differences between real and generated distributions. To address this limitation, we propose Adversarial Fréchet Distance (AdvFD), which complements the static representation targets in FD-Loss with a calibrated adversarially learned representation. AdvFD augments the original static Fréchet objective with a learnable representation that adversarially maximizes the Fréchet discrepancy between real and generated samples, while the generator minimizes the same discrepancy in the resulting adaptive feature space. To prevent the adversarial representation from trivially increasing the objective through feature amplification, we further introduce real-feature whitening, which normalizes its scale and covariance geometry and stabilizes the min--max optimization. Extensive experiments show that AdvFD consistently improves one-step generator post-training across both JiT and pMF backbones and across different model scales.

Mingju Gao, Jingkai Zhou, Kun Gai +2

Peking University · KlingAI Research

Diffusion Model Alignment Diffusion Model Fine-Tuning Diffusion-Based Image Generation

Aug 9, 2026cs.CV

High-Quality Exposure Correction with Diffusion-Based Image Generation Priors

Although most existing exposure correction methods achieve high fidelity, they often place excessive focus on overall pixel-wise accuracy, making it challenging to effectively model extreme exposure regions, which results in suboptimal perceptual quality. Recently, diffusion models have received significant attention due to their remarkable performance in the realm of image generation. However, their successful application to exposure correction remains a challenging and open question. The key challenge lies in generating accurate image structures and maintaining high image fidelity during stochastic diffusion processes. In this paper, we propose DPEC (Diffusion Prior-based Exposure Correction), a novel framework for image exposure correction that utilizes diffusion-based image generation priors encapsulated in pre-trained large-scale diffusion models. Specifically, we first propose an efficient fine-tuning strategy to derive an exposure corrector from pre-trained models, enabling the generation of enhanced images in a single-step denoising process. Moreover, we seamlessly combine the strengths of diffusion models and regression models, and design a joint cross-attention module to integrate multi-scale diffusion prior features, thereby effectively preserving high-frequency details and minimizing random artifacts. The diffusion model focuses on dealing with low-frequency content rather than all the intricate texture details. The experimental results demonstrate that the proposed DPEC method consistently outperforms existing state-of-the-art methods on multiple exposure correction datasets, whether in terms of fidelity, perceptual quality, or visual effects.

Ziwen Li, Meng Cao, Jinpu Zhang +4

National Key Laboratory of Science and Technology on Multispectral Information Processing, School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan 430074, China · Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE · National University of Defense Technology, Changsha, China +1

Image Enhancement Diffusion Model Fine-Tuning

Aug 7, 2026cs.CV

PAST: Prompt-Adaptive Sampling Termination for Efficient Diffusion Model

While diffusion models have made significant progress in text-to-image tasks, they still exhibit limitations when directly optimizing downstream objectives. Although Reinforcement Learning (RL) enables targeted optimization, existing methods are generally constrained by low-efficiency fine-tuning and sparse rewards. To address these challenges, we propose PAST, which provides differentiated rewards while adaptively regulating training episode length by jointly perceiving denoising progress and prompt difficulty. Specifically, we design an intrinsic reward paradigm to compensate for sparse extrinsic rewards and guide the model to explore paths that diverge more efficiently from noise patterns. We further provide theoretical justification for intrinsic rewards. Then, PAST dynamically monitors denoising completion and semantic alignment between image structures and prompt semantics. When both metrics satisfy generation requirements, the system adaptively terminates training. This enables appropriate allocation of episode lengths based on prompt difficulty and the current generation process. Finally, based on the predicted residual noise level, we establish a dual adaptive coordination mechanism. Specifically, it not only balances the extrinsic and intrinsic rewards but also balances the exploration and convergence. Experimental results demonstrate that PAST enhances computational efficiency of existing RL fine-tuning methods by up to 66.7%, while improving preference optimization quality by up to 29.5% through its dual adaptive regulation mechanism.

Renye Yan, Jikang Cheng, You Wu +4

Peking University Beijing, China

Intrinsic Reward Methods Reinforcement Learning Diffusion Sampling Diffusion Model Fine-Tuning RL for Diffusion Models

Aug 6, 2026cs.CV

Sample-Adaptive Latent Rewards for Uncertainty-Guided Diffusion Post-Training

Latent reward models can supervise visual diffusion models without decoding intermediate states into pixel space. This makes alignment with human preferences more efficient. However, existing latent reward models output only scalar scores. They do not estimate the uncertainty of each prediction. The generator therefore cannot determine which feedback is reliable. This can drive optimization in the wrong direction and lead to reward hacking. We propose \textsc{SURE}, a unified latent-space framework for image and video diffusion models. It learns reward distributions and directly uses their reliability to guide dense post-training. First, we propose sample-adaptive latent reward model (\textsc{SURE-LRM}). It predicts a Gaussian utility for each noisy latent. Its mean predicts the reward score. Its variance reflect the uncertainty of prediction without human annotation. The learned distribution then guides post-training through uncertainty-guided reward feedback learning (\textsc{SURE-REFL}). This method provides uncertainty-guided dense feedback along the denoising trajectory. At selected transitions, \textsc{SURE-REFL} queries the frozen \textsc{SURE-LRM}. It converts detached variance into reliability weights for samples at the same transition. Each weighted reward is backpropagated only through its local transition. The entire process remains in latent space and requires neither pixel-space decoding nor the full denoising graph. Experiments show that \textsc{SURE-LRM} improves preference prediction over strong baselines. \textsc{SURE-REFL} achieves the sota performance among various metrics and further improves optimization stability. It also achieves the highest VBench quality, semantic, and total scores among the evaluated methods.

Rui Li, Yuanzhi Liang, Ke Hao +4

University of Science and Technology of China, Hefei, China · Institute of Artificial Intelligence, China Telecom (TeleAI), Shanghai, China · Shanghai Jiao Tong University, Shanghai, China

Diffusion Model Alignment Video Diffusion Models Diffusion Model Fine-Tuning Diffusion Models

Aug 5, 2026cs.CV

When does training on downscaled images yield the same gradients?

Diffusion transformers deliver strong image generation, but their training cost grows superlinearly with resolution. Recent work justifies training or sampling at reduced resolution on a spectral premise: at high noise, a downscaled latent preserves almost the full surviving signal. Whether a downscaled step also preserves the native training gradient signal, however, has remained unresolved. We reduce how that signal changes under downscaling to two terms: a noise-dependent term governed by the downscale ratio, which decays at high noise as the spectral premise predicts, and a σ-independent floor governed by the target grid's absolute token count, carried by the compute graph itself and removed by no noise level. The measured (route, σ) map corroborates the account and uncovers structure the spectral picture cannot express: on the 1024->768 route, a window (0.65 < σ< 0.95), predicted by no spectral criterion at any tolerance, where the downscaled gradient stays within a small margin of the native one. Training LoRA adapters with downscaled steps restricted to the routes and noise windows the map validates reduces training time by 14.6% at a fixed step budget while remaining near-native in weight space. Code is available at https://github.com/sorryhyun/anima_lora.

Seunghyun Ji

Independent researcher

Diffusion Transformer Diffusion Model Fine-Tuning Diffusion Model Acceleration Diffusion Models