RLHF techniques like DPO can significantly improve the generation quality of text-to-image diffusion models. However, these methods optimize for a single reward that aligns model generation with population-level preferences, neglecting the nuances of individual users’ beliefs or values. This lack of personalization limits the efficacy of these models. To bridge this gap, we introduce PPD, a multi-reward optimization objective that aligns diffusion models with personalized preferences. With PPD, a diffusion model learns the individual preferences of a population of users in a few-shot way, enabling generalization to unseen users. Specifically, our approach (1) leverages a vision-language model (VLM) to extract personal preference embeddings from a small set of pairwise preference examples, and then (2) incorporates the embeddings into diffusion models through cross attention. Conditioning on user embeddings, the text-to-image models are fine-tuned with the DPO objective, simultaneously optimizing for alignment with the preferences of multiple users. Empirical results demonstrate that our method effectively optimizes for multiple reward functions and can interpolate between them during inference. In real-world user scenarios, with as few as four preference examples from a new user, our approach achieves an average win rate of 76% over Stable Cascade, generating images that more accurately reflect specific user preferences.

Diffusion models and Flow Matching generate high-quality samples but are slow at inference, and distilling them into few-step models often leads to instability and extensive tuning. To resolve these trade-offs, we propose Moment Matching Self-Distillation (MMSD), a new class of generative models for one- or few-step sampling with a single-stage training procedure. Unlike distillation, MMSD does not require pre-training initialization and optimization of two networks; and unlike Consistency Models, MMSD guarantees distribution-level convergence and remains stable under various hyperparameters and standard model architectures. MMSD surpasses diffusion models on ImageNet-256x256 with 2.13 FID using only 8 inference steps and achieves state-of-the-art 2-step FID of 2.05 on CIFAR-10 for a model trained from scratch.

Large language models (LLMs) are fine-tuned using human comparison data with Reinforcement Learning from Human Feedback (RLHF) methods to make them better aligned with users' preferences. In contrast to LLMs, human preference learning has not been widely explored in text-to-image diffusion models; the best existing approach is to fine-tune a pretrained model using carefully curated high quality images and captions to improve visual appeal and text alignment. We propose Diffusion-DPO, a method to align diffusion models to human preferences by directly optimizing on human comparison data. Diffusion-DPO is adapted from the recently developed Direct Preference Optimization (DPO) [36], a simpler alternative to RLHF which directly optimizes a policy that best satisfies human preferences under a classification objective. We re-formulate DPO to account for a diffusion model notion of likelihood, utilizing the evidence lower bound to derive a differentiable objective. Using the Pick-a-Pic dataset of 851K crowdsourced pairwise preferences, we fine-tune the base model of the state-of-the-art Stable Diffusion XL (SDXL)-1.0 model with Diffusion-DPO. Our fine-tuned base model significantly outperforms both base SDXL-1.0 and the larger SDXL-1.0 model consisting of an additional refinement model in human evaluation, improving visual appeal and prompt alignment. We also develop a variant that uses AI feedback and has comparable performance to training on human preferences, opening the door for scaling of diffusion model alignment methods.

Recent methods such as Score Distillation Sampling (SDS) and Variational Score Distillation (VSD) using 2D diffusion models for text-to-3D generation have demon-strated impressive generation quality. However, the long generation time of such algorithms significantly degrades the user experience. To tackle this problem, we propose DreamPropeller, a drop-in acceleration algorithm that can be wrapped around any existing text-to-3D generation pipeline based on score distillation. Our framework generalizes Picard iterations, a classical algorithm for parallel sampling an ODE path, and can account for non-ODE paths such as momentum-based gradient updates and changes in dimensions during the optimization process as in many cases of 3D generation. We show that our algorithm trades parallel compute for wallclock time and empirically achieves up to 4.7x speedup with a negligible drop in generation quality for all tested frameworks. Our implementation can be found here.

Diffusion models have achieved state-of-the-art results on many modalities including images, speech, and video. However, existing models are not tailored to support remote sensing data, which is widely used in important applications including environmental monitoring and crop-yield prediction. Satellite images are significantly different from natural images -- they can be multi-spectral, irregularly sampled across time -- and existing diffusion models trained on images from the Web do not support them. Furthermore, remote sensing data is inherently spatio-temporal, requiring conditional generation tasks not supported by traditional methods based on captions or images. In this paper, we present DiffusionSat, to date the largest generative foundation model trained on a collection of publicly available large, high-resolution remote sensing datasets . As text-based captions are sparsely available for satellite images, we incorporate the associated metadata such as geolocation as conditioning information. Our method produces realistic samples and can be used to solve multiple generative tasks including temporal generation, multi-spectral superrresolution and in-painting. Our method outperforms previous state-of-the-art methods for satellite image generation and is the first large-scale _generative_ foundation model for satellite imagery. The project website can be found here: https://samar-khanna.github.io/DiffusionSat/

Diffusion models are powerful generative models that map noise to data using stochastic processes. However, for many applications such as image editing, the model input comes from a distribution that is not random noise. As such, diffusion models must rely on cumbersome methods like guidance or projected sampling to incorporate this information in the generative process. In our work, we propose Denoising Diffusion Bridge Models (DDBMs), a natural alternative to this paradigm based on *diffusion bridges*, a family of processes that interpolate between two paired distributions given as endpoints. Our method learns the score of the diffusion bridge from data and maps from one endpoint distribution to the other by solving a (stochastic) differential equation based on the learned score. Our method naturally unifies several classes of generative models, such as score-based diffusion models and OT-Flow-Matching, allowing us to adapt existing design and architectural choices to our more general problem. Empirically, we apply DDBMs to challenging image datasets in both pixel and latent space. On standard image translation problems, DDBMs achieve significant improvement over baseline methods, and, when we reduce the problem to image generation by setting the source distribution to random noise, DDBMs achieve comparable FID scores to state-of-the-art methods despite being built for a more general task.

We introduce AO-Grasp, a grasp proposal method that generates 6 DoF grasps that enable robots to interact with articulated objects, such as opening and closing cabinets and appliances. AO-Grasp consists of two main contributions: the AO-Grasp Model and the AO-Grasp Dataset. Given a segmented partial point cloud of a single articulated object, the AO-Grasp Model predicts the best grasp points on the object with an Actionable Grasp Point Predictor. Then, it finds corresponding grasp orientations for each of these points, resulting in stable and actionable grasp proposals. We train the AO-Grasp Model on our new AO-Grasp Dataset, which contains 78K actionable parallel-jaw grasps on synthetic articulated objects. In simulation, AO-Grasp achieves a 45.0% grasp success rate, whereas the highest performing baseline achieves a 35.0% success rate. Additionally, we evaluate AO-Grasp on 120 real-world scenes of objects with varied geometries, articulation axes, and joint states, where AO-Grasp produces successful grasps on 67.5% of scenes, while the baseline only produces successful grasps on 33.3% of scenes. To the best of our knowledge, AO-Grasp is the first method for generating 6 DoF grasps on articulated objects directly from partial point clouds without requiring part detection or hand-designed grasp heuristics. The AO-Grasp Dataset and a pre-trained AO-Grasp model are available at our project website: https://stanford-iprl-lab.github.io/ao-grasp/.

Agents must be able to adapt quickly as an environment changes. We find that existing model-based reinforcement learning agents are unable to do this well, in part because of how they use past experiences to train their world model. Here, we present Curious Replay—a form of prioritized experience replay tailored to model-based agents through use of a curiosity-based priority signal. Agents using Curious Replay exhibit improved performance in an exploration paradigm inspired by animal behavior and on the Crafter benchmark. DreamerV3 with Curious Replay surpasses state-of-the-art performance on Crafter, achieving a mean score of 19.4 that substantially improves on the previous high score of 14.5 by DreamerV3 with uniform replay, while also maintaining similar performance on the Deepmind Control Suite. Code for Curious Replay is available at github.com/AutonomousAgentsLab/curiousreplay.

Methods based on ordinary differential equations (ODEs) are widely used to build generative models of time-series. In addition to high computational overhead due to explicitly computing hidden states recurrence, existing ODE-based models fall short in learning sequence data with sharp transitions - common in many real-world systems - due to numerical challenges during optimization. In this work, we propose LS4, a generative model for sequences with latent variables evolving according to a state space ODE to increase modeling capacity. Inspired by recent deep state space models (S4), we achieve speedups by leveraging a convolutional representation of LS4 which bypasses the explicit evaluation of hidden states. We show that LS4 significantly outperforms previous continuous-time generative models in terms of marginal distribution, classification, and prediction scores on real-world datasets in the Monash Forecasting Repository, and is capable of modeling highly stochastic data with sharp temporal transitions. LS4 sets state-of-the-art for continuous-time latent generative models, with significant improvement of mean squared error and tighter variational lower bounds on irregularly-sampled datasets, while also being x100 faster than other baselines on long sequences.

Normalizing flows model complex probability distributions using maps obtained by composing invertible layers. Special linear layers such as masked and 1{\texttimes}1 convolutions play a key role in existing architectures because they increase expressive power while having tractable Jacobians and inverses. We propose a new family of invertible linear layers based on butterfly layers, which are known to theoretically capture complex linear structures including permutations and periodicity, yet can be inverted efficiently. This representational power is a key advantage of our approach, as such structures are common in many real-world datasets. Based on our invertible butterfly layers, we construct a new class of normalizing flow mod- els called ButterflyFlow. Empirically, we demonstrate that ButterflyFlows not only achieve strong density estimation results on natural images such as MNIST, CIFAR-10, and ImageNet-32{\texttimes}32, but also obtain significantly better log-likelihoods on structured datasets such as galaxy images and MIMIC-III patient cohorts{—}all while being more efficient in terms of memory and computation than relevant baselines.

We propose a novel approach for probabilistic generative modeling of 3D shapes. Unlike most existing models that learn to deterministically translate a latent vector to a shape, our model, Point-Voxel Diffusion (PVD), is a unified, probabilistic formulation for unconditional shape generation and conditional, multi-modal shape completion. PVD marries denoising diffusion models with the hybrid, point-voxel representation of 3D shapes. It can be viewed as a series of denoising steps, reversing the diffusion process from observed point cloud data to Gaussian noise, and is trained by optimizing a variational lower bound to the (conditional) likelihood function. Experiments demonstrate that PVD is capable of synthesizing high-fidelity shapes, completing partial point clouds, and generating multiple completion results from single-view depth scans of real objects.

Open-domain dialogue generation has gained increasing attention in Natural Language Processing.Its evaluation requires a holistic means.Human ratings are deemed as the gold standard.As human evaluation is inefficient and costly, an automated substitute is highly desirable.In this paper, we propose holistic evaluation metrics that capture different aspects of open-domain dialogues.Our metrics consist of (1) GPT-2 based context coherence between sentences in a dialogue, (2) GPT-2 based fluency in phrasing, (3) n-gram based diversity in responses to augmented queries, and (4) textual-entailment-inference based logical self-consistency.The empirical validity of our metrics is demonstrated by strong correlations with human judgments.We open source the code and relevant materials.1

This paper proposes a joint training method to learn both the variational auto-encoder (VAE) and the latent energy-based model (EBM). The joint training of VAE and latent EBM are based on an objective function that consists of three Kullback-Leibler divergences between three joint distributions on the latent vector and the image, and the objective function is of an elegant symmetric and anti-symmetric form of divergence triangle that seamlessly integrates variational and adversarial learning. In this joint training scheme, the latent EBM serves as a critic of the generator model, while the generator model and the inference model in VAE serve as the approximate synthesis sampler and inference sampler of the latent EBM. Our experiments show that the joint training greatly improves the synthesis quality of the VAE. It also enables learning of an energy function that is capable of detecting out of sample examples for anomaly detection.