NeurIPS 2025 论文列表

We propose ReinFlow, a simple yet effective online reinforcement learning (RL) framework that fine-tunes a family of flow matching policies for continuous robotic control. Derived from rigorous RL theory, ReinFlow injects learnable noise into a flow policy’s deterministic path, converting the flow into a discrete-time Markov Process for exact and straightforward likelihood computation. This conversion facilitates exploration and ensures training stability, enabling ReinFlow to fine-tune diverse flow model variants stably, including Rectified Flow [34] and Shortcut Models [18], particularly at very few or even one denoising step. We benchmark ReinFlow in representative locomotion and manipulation tasks, including long- horizon planning with visual input and sparse reward. The episode reward of Rectified Flow policies obtained an average net growth of 135.36% after fine-tuning in challenging legged locomotion tasks while saving denoising steps and 82.63% of wall time compared to state-of-the-art diffusion RL fine-tuning method DPPO [42]. The success rate of the Shortcut Model policies in state and visual manipulation tasks achieved an average net increase of 40.34% after fine-tuning with ReinFlow at four or even one denoising step, whose performance is comparable to fine-tuned DDIM policies while saving computation time for an average of 23.20% . Code, model, and checkpoints available on the project website: https://reinflow.github.io/

Recent advances in diffusion models have dramatically improved image fidelity and diversity. However, aligning these models with nuanced human preferences -such as aesthetics, engagement, and subjective appeal remains a key challenge due to the scarcity of large-scale human annotations. Collecting such data is both expensive and limited in diversity. To address this, we leverage the reasoning capabilities of vision-language models (VLMs) and propose Self-Play Reward Optimization (SPRO), a scalable, annotation-free training framework based on multimodal self-play. SPRO learns to jointly align prompt and image generation with human preferences by iteratively generating, evaluating, and learning to refine outputs using synthetic reward signals such as aesthetics and human engagement. This self-improving feedback loop eliminates the need for external supervision. SPRO comprises three stages: (1) SPRO-Prompt, which trains a Guider-VLM via self-play to generate diverse, high-reward prompts targeting objectives such as PickScore (user preference), LAION-Aesthetics, and EngageNet (engagement); (2) SPRO-Image, which fine-tunes the diffusion model on high-reward images derived from these prompts; and (3) SPRO-Multimodal (SPRO-MM), which integrates both components for full end-to-end alignment. Without relying on human-labeled data, SPRO achieves an average 30\% improvement across preference objectives. Moreover, its generated prompts generalize across both open- and closed-source diffusion models. Through iterative self-play, SPRO discovers prompting strategies rarely authored by humans such as emphasizing visual harmony for aesthetics or leveraging shadow-based cues for engagement. SPRO offers a scalable path toward aligning generative models with complex subjective human values.

We propose a framework for learning maps between probability distributions that broadly generalizes the time dynamics of flow and diffusion models. To enable this, we generalize stochastic interpolants by replacing the scalar time variable with vectors, matrices, or linear operators, allowing us to bridge probability distributions across multiple dimensional spaces. This approach enables the construction of versatile generative models capable of fulfilling multiple tasks without task-specific training. Our operator-based interpolants not only provide a unifying theoretical perspective for existing generative models but also extend their capabilities. Through numerical experiments, we demonstrate the zero-shot efficacy of our method on conditional generation and inpainting, fine-tuning and posterior sampling, and multiscale modeling, suggesting its potential as a generic task-agnostic alternative to specialized models.

Diffusion bridge models and stochastic interpolants enable high-quality image-to-image (I2I) translation by creating paths between distributions in pixel space. However, recent diffusion bridge models excel in image translation but suffer from restricted design flexibility and complicated hyperparameter tuning, whereas Stochastic Interpolants offer greater flexibility but lack essential refinements. We show that these complementary strengths can be unified by interpreting all existing methods within a single SI-based framework. In this work, we unify and expand the space of bridge models by extending Stochastic Interpolants (SIs) with preconditioning, endpoint conditioning, and an optimized sampling algorithm. These enhancements expand the design space of diffusion bridge models, leading to state-of-the-art performance in both image quality and sampling efficiency across diverse I2I tasks. Furthermore, we identify and address a previously overlooked issue of low sample diversity under fixed conditions. We introduce a quantitative analysis for output diversity and demonstrate how we can modify the base distribution for further improvements.

Recent work shows that features from Stable Diffusion (SD) and contrastively pretrained models like DINO can be directly used for zero-shot semantic correspondence via naive feature concatenation. In this paper, we explore the stronger potential of Stable Diffusion 3 (SD3), a rectified flow-based model with a multimodal transformer backbone (MM-DiT). We show that semantic signals in SD3 are scattered across multiple timesteps and transformer layers, and propose a multi-level fusion scheme to extract discriminative features. Moreover, we identify that naive fusion across models suffers from inconsistent distributions, thus leading to suboptimal performance. To address this, we propose a simple yet effective confidence-aware feature fusion strategy that re-weights each model’s contribution based on prediction confidence scores derived from their matching uncertainties. Notably, this fusion approach is not only training-free but also enables per-pixel adaptive integration of heterogeneous features. The resulting representation, Fuse2Match, significantly outperforms strong baselines on SPair-71k, PF-Pascal, and PSC6K, validating the benefit of combining SD3, SD, and DINO through our proposed confidence-aware feature fusion. Code is available at https://github.com/panda7777777/fuse2match

Lip synchronization is the task of aligning a speaker’s lip movements in video with corresponding speech audio, and it is essential for creating realistic, expressive video content. However, existing methods often rely on reference frames and masked-frame inpainting, which limit their robustness to identity consistency, pose variations, facial occlusions, and stylized content. In addition, since audio signals provide weaker conditioning than visual cues, lip shape leakage from the original video will affect lip sync quality. In this paper, we present OmniSync, a universal lip synchronization framework for diverse visual scenarios. Our approach introduces a mask-free training paradigm using Diffusion Transformer models for direct frame editing without explicit masks, enabling unlimited-duration inference while maintaining natural facial dynamics and preserving character identity. During inference, we propose a flow-matching-based progressive noise initialization to ensure pose and identity consistency, while allowing precise mouth-region editing. To address the weak conditioning signal of audio, we develop a Dynamic Spatiotemporal Classifier-Free Guidance (DS-CFG) mechanism that adaptively adjusts guidance strength over time and space. We also establish the AIGC-LipSync Benchmark, the first evaluation suite for lip synchronization in diverse AI-generated videos. Extensive experiments demonstrate that OmniSync significantly outperforms prior methods in both visual quality and lip sync accuracy, achieving superior results in both real-world and AI-generated videos.

While diffusion models have shown promise for combinatorial optimization (CO), their inference-time scaling cost-efficiency remains relatively underexplored. Existing methods improve solution quality by increasing denoising steps, but the performance often becomes saturated quickly. This paper proposes GenSCO to systematically scale diffusion solvers by an orthogonal dimension of inference-time computation beyond denoising step expansion, i.e., search-driven generation. GenSCO takes generation as a search operator rather than a complete solving process, where each operator cycle combines solution disruption (via local search operators) and diffusion sampling, enabling iterative exploration of the learned solution space. Rather than over-refining current solutions, this paradigm encourages the model to leave local optima and explore a broader area of the solution space, ensuring a more consistent scaling effect. The search loop is supported by a search-friendly solution-enhancement training procedure that incorporates a rectified flow model learning to establish diffusion trajectories between suboptimal solutions and the optimal ones. The flow model is empowered by a lightweight transformer architecture to learn neural ODEs that linearize solution trajectories, accelerating convergence of the scaling effect with efficiency. The resulting enhanced scaling efficiency and practical scalability lead to synergistic performance improvements. Extensive experiments show that GenSCO delivers performance improvements by orders of magnitude over previous state-of-the-art neural methods. Notably, GenSCO even achieves significant speedups compared to the state-of-the-art classic mathematical solver LKH3, delivering a 141x speedup to reach 0.000% optimality gap on TSP-100, and approximately a 10x speedup to reach 0.02% on TSP-500.

Geometric diffusion models have shown remarkable success in molecular dynamics and structure generation. However, efficiently fine-tuning them for downstream tasks with varying geometric controls remains underexplored. In this work, we propose an SE(3)-equivariant adapter framework (GeoAda) that enables flexible and parameter-efficient fine-tuning for controlled generative tasks without modifying the original model architecture. GeoAda introduces a structured adapter design: control signals are first encoded through coupling operators, then processed by a trainable copy of selected base model layers, and finally projected back via decoupling operators followed by an equivariant zero-initialized convolution. By fine-tuning only these lightweight adapter modules, GeoAda preserves the model’s geometric consistency while mitigating overfitting and catastrophic forgetting. We theoretically prove that the proposed adapters maintain SE(3)-equivariance, ensuring that the geometric inductive biases of the pretrained diffusion model remain intact during adaptation. We demonstrate the wide applicability of \method across diverse geometric control types, including frame control, global control, subgraph control, and a broad range of application domains such as particle dynamics, molecular dynamics, human motion prediction, and molecule generation. Empirical results show that GeoAda achieves state-of-the-art fine-tuning performance while preserving original task accuracy, whereas other baselines experience significant performance degradation due to overfitting and catastrophic forgetting.

Current text-to-image (T2I) generation models struggle to align spatial composition with the input text, especially in complex scenes. Even layout-based approaches yield suboptimal spatial control, as their generation process is decoupled from layout planning, making it difficult to refine the layout during synthesis. We present CoT-Diff, a framework that brings step-by-step CoT-style reasoning into T2I generation by tightly integrating Multimodal Large Language Model (MLLM)-driven 3D layout planning with the diffusion process. CoT-Diff enables layout-aware reasoning inline within a single diffusion round: at each denoising step, the MLLM evaluates intermediate predictions, dynamically updates the 3D scene layout, and continuously guides the generation process. The updated layout is converted into semantic conditions and depth maps, which are fused into the diffusion model via a condition-aware attention mechanism, enabling precise spatial control and semantic injection. Experiments on 3D Scene benchmarks show that CoT-Diff significantly improves spatial alignment and compositional fidelity, and outperforms the state-of-the-art method by 34.7% in complex scene spatial accuracy, thereby validating the effectiveness of this entangled generation paradigm.

While Classifier-Free Guidance (CFG) has become standard for improving sample fidelity in conditional diffusion models, it can harm diversity and induce memorization by applying constant guidance regardless of whether a particular sample needs correction. We propose **F**eed**B**ack **G**uidance (FBG), which uses a state-dependent coefficient to self-regulate guidance amounts based on need. Our approach is derived from first principles by assuming the learned conditional distribution is linearly corrupted by the unconditional distribution, contrasting with CFG's implicit multiplicative assumption. Our scheme relies on feedback of its own predictions about the conditional signal informativeness to adapt guidance dynamically during inference, challenging the view of guidance as a fixed hyperparameter. The approach is benchmarked on ImageNet512x512, where it significantly outperforms Classifier-Free Guidance and is competitive to Limited Interval Guidance (LIG) while benefitting from a strong mathematical framework. On Text-To-Image generation, we demonstrate that, as anticipated, our approach automatically applies higher guidance scales for complex prompts than for simpler ones and that it can be easily combined with existing guidance schemes such as CFG or LIG.

Diffusion transformers (DiTs) struggle to generate images at resolutions higher than their training resolutions. The primary obstacle is that the explicit positional encodings (PE), such as RoPE, need extrapolating to unseen positions which degrades performance when the inference resolution differs from training. In this paper, We propose a Length-Extrapolatable Diffusion Transformer (LEDiT) to overcome this limitation. LEDiT needs no explicit PEs, thereby avoiding PE extrapolation. The key innovation of LEDiT lies in the use of causal attention. We demonstrate that causal attention can implicitly encode global positional information and show that such information facilitates extrapolation. We further introduce a locality enhancement module, which captures fine-grained local information to complement the global coarse-grained position information encoded by causal attention. Experimental results on both conditional and text-to-image generation tasks demonstrate that LEDiT supports up to 4× resolution scaling (e.g., from 256$\times$256 to 512$\times$512), achieving better image quality compared to the state-of-the-art length extrapolation methods. We believe that LEDiT marks a departure from the standard RoPE-based methods and offers a promising insight into length extrapolation. Project page: https://shenzhang2145.github.io/ledit/

We propose RoPECraft, a training-free video motion transfer method for diffusion transformers that operates solely by modifying their rotary positional embeddings (RoPE). We first extract dense optical flow from a reference video, and utilize the resulting motion offsets to warp the complex-exponential tensors of RoPE, effectively encoding motion into the generation process. These embeddings are then further optimized during denoising time steps via trajectory alignment between the predicted and target velocities using a flow-matching objective. To keep the output faithful to the text prompt and prevent duplicate generations, we incorporate a regularization term based on the phase components of the reference video’s Fourier transform, projecting the phase angles onto a smooth manifold to suppress high-frequency artifacts. Experiments on benchmarks reveal that RoPECraft outperforms all recently published methods, both qualitatively and quantitatively.

A cascade over a network refers to the diffusion process where behavior changes occurring in one part of an interconnected population lead to a series of sequential changes throughout the entire population. In recent years, there has been a surge in interest and efforts to understand and model cascade mechanisms since they motivate many significant research topics across different disciplines. The propagation structure of cascades is governed by underlying diffusion networks that are often hidden. Inferring diffusion networks thus enables interventions in cascading process to maximize information propagation and provides insights into the Granger causality of interaction mechanisms among individuals. In this project, we propose a novel double network mixture model for inferring latent diffusion network in presence of strong cascade heterogeneity. The new model represents cascade pathways as a distributional mixture over diffusion networks that capture different cascading patterns at the population level. We develop a data-driven optimization method to infer diffusion networks using only visible temporal cascade records, avoiding the need to model complex and heterogeneous individual states. Both statistical and computational guarantees are established for the proposed method. We apply the proposed model to analyze research topic cascades in social sciences across U.S. universities and uncover the latent research topic diffusion network among top U.S. social science programs.

Real-world datasets collected from sensors or human inputs are prone to noise and errors, posing significant challenges for applying offline reinforcement learning (RL). While existing methods have made progress in addressing corrupted actions and rewards, they remain insufficient for handling corruption in high-dimensional state spaces and for cases where multiple elements in the dataset are corrupted simultaneously. Diffusion models, known for their strong denoising capabilities, offer a promising direction for this problem—but their tendency to overfit noisy samples limits their direct applicability. To overcome this, we propose **A**mbient **D**iffusion-**G**uided Dataset Recovery (**ADG**), a novel approach that pioneers the use of diffusion models to tackle data corruption in offline RL. First, we introduce Ambient Denoising Diffusion Probabilistic Models (DDPM) from approximated distributions, which enable learning on partially corrupted datasets with theoretical guarantees. Second, we use the noise-prediction property of Ambient DDPM to distinguish between clean and corrupted data, and then use the clean subset to train a standard DDPM. Third, we employ the trained standard DDPM to refine the previously identified corrupted data, enhancing data quality for subsequent offline RL training. A notable strength of ADG is its versatility—it can be seamlessly integrated with any offline RL algorithm. Experiments on a range of benchmarks, including MuJoCo, Kitchen, and Adroit, demonstrate that ADG effectively mitigates the impact of corrupted data and improves the robustness of offline RL under various noise settings, achieving state-of-the-art results.

We present a systematic theoretical framework that interprets masked diffusion models (MDMs) as solutions to energy minimization problems in discrete optimal transport. Specifically, we prove that three distinct energy formulations—kinetic, conditional kinetic, and geodesic energy—are mathematically equivalent under the structure of MDMs, and that MDMs minimize all three when the mask schedule satisfies a closed-form optimality condition. This unification not only clarifies the theoretical foundations of MDMs, but also motivates practical improvements in sampling. By parameterizing interpolation schedules via Beta distributions, we reduce the schedule design space to a tractable 2D search, enabling efficient post-training tuning without model modification. Experiments on synthetic and real-world benchmarks demonstrate that our energy-inspired schedules outperform hand-crafted baselines, particularly in low-step sampling settings.

Generative modeling, representation learning, and classification are three core problems in machine learning (ML), yet their state-of-the-art (SoTA) solutions remain largely disjoint. In this paper, we ask: Can a unified principle address all three? Such unification could simplify ML pipelines and foster greater synergy across tasks. We introduce Latent Zoning Network (LZN) as a step toward this goal. At its core, LZN creates a shared Gaussian latent space that encodes information across all tasks. Each data type (e.g., images, text, labels) is equipped with an encoder that maps samples to disjoint latent zones, and a decoder that maps latents back to data. ML tasks are expressed as compositions of these encoders and decoders: for example, label-conditional image generation uses a label encoder and image decoder; image embedding uses an image encoder; classification uses an image encoder and label decoder. We demonstrate the promise of LZN in three increasingly complex scenarios: (1) LZN can enhance existing models (image generation): When combined with the SoTA Rectified Flow model, LZN improves FID on CIFAR10 from 2.76 to 2.59—without modifying the training objective. (2) LZN can solve tasks independently (representation learning): LZN can implement unsupervised representation learning without auxiliary loss functions, outperforming the seminal MoCo and SimCLR methods by 9.3% and 0.2%, respectively, on downstream linear classification on ImageNet. (3) LZN can solve multiple tasks simultaneously (joint generation and classification): With image and label encoders/decoders, LZN performs both tasks jointly by design, improving FID and achieving SoTA classification accuracy on CIFAR10. The code and trained models are available at https://github.com/microsoft/latent-zoning-networks. The project website is at https://zinanlin.me/blogs/latent_zoning_networks.html.

Modeling the transport dynamics of natural processes from population-level observations is a ubiquitous problem in the natural sciences. Such models rely on key assumptions about the underlying process in order to enable faithful learning of governing dynamics that mimic the actual system behavior. The de facto assumption in current approaches relies on the principle of least action that results in gradient field dynamics and leads to trajectories minimizing an energy functional between two probability measures. However, many real-world systems, such as cell cycles in single-cell RNA, are known to exhibit non-gradient, periodic behavior, which fundamentally cannot be captured by current state-of-the-art methods such as flow and bridge matching. In this paper, we introduce Curly Flow Matching (Curly-FM), a novel approach that is capable of learning non-gradient field dynamics by designing and solving a Schrödinger bridge problem with a non-zero drift reference process---in stark contrast to typical zero-drift reference processes---which is constructed using inferred velocities in addition to population snapshot data. We showcase Curly-FM by solving the trajectory inference problems for single cells, computational fluid dynamics, and ocean currents with approximate velocities. We demonstrate that Curly-FM can learn trajectories that better match both the reference process and population marginals. Curly-FM expands flow matching models beyond the modeling of populations and towards the modeling of known periodic behavior in physical systems. Our code repository is accessible at: https://github.com/kpetrovicc/curly-flow-matching.git

Incorporating pre-collected offline data from a source environment can significantly improve the sample efficiency of reinforcement learning (RL), but this benefit is often challenged by discrepancies between the transition dynamics of the source and target environments. Existing methods typically address this issue by penalizing or filtering out source transitions in high dynamics-gap regions. However, their estimation of the dynamics gap often relies on KL divergence or mutual information, which can be ill-defined when the source and target dynamics have disjoint support. To overcome these limitations, we propose CompFlow, a method grounded in the theoretical connection between flow matching and optimal transport. Specifically, we model the target dynamics as a conditional flow built upon the output distribution of the source-domain flow, rather than learning it directly from a Gaussian prior. This composite structure offers two key advantages: (1) improved generalization for learning target dynamics, and (2) a principled estimation of the dynamics gap via the Wasserstein distance between source and target transitions. Leveraging our principled estimation of the dynamics gap, we further introduce an optimistic active data collection strategy that prioritizes exploration in regions of high dynamics gap, and theoretically prove that it reduces the performance disparity with the optimal policy. Empirically, CompFlow outperforms strong baselines across several RL benchmarks with shifted dynamics.

Recent large language models (LLMs) have demonstrated strong reasoning capabilities that benefits from online reinforcement learning (RL). These capabilities have primarily been demonstrated within the left-to-right autoregressive (AR) generation paradigm. In contrast, non-autoregressive paradigms based on diffusion generate text in a coarse-to-fine manner. Although recent diffusion-based large language models (dLLMs) have achieved competitive language modeling performance compared to their AR counterparts, it remains unclear if dLLMs can also leverage recent advances in LLM reasoning. To this end, we propose, a framework to adapt pre-trained masked dLLMs into reasoning models via a combination of supervised finetuning (SFT) and RL. Specifically, we develop and extend techniques to improve reasoning in pretrained dLLMs: (a) we utilize a masked SFT technique to distill knowledge and instill self-improvement behavior directly from existing datasets, and (b) we introduce a novel critic-free, policy-gradient based RL algorithm called diffu-GRPO, the first integration of policy gradient methods to masked dLLMs. Through empirical studies, we investigate the performance of different post-training recipes on multiple mathematical and planning benchmarks. We find that d1 yields the best performance and significantly improves performance of a state-of-the-art dLLM.