CDM migrates distribution matching distillation to continuous time via dynamic random-length schedules and active off-trajectory latent alignment, yielding competitive few-step image fidelity on SD3 and Longcat-Image.
hub
Score-based generative modeling through stochastic differential equations
27 Pith papers cite this work. Polarity classification is still indexing.
27
Pith papers citing it
hub tools
citation-role summary
background 3
citation-polarity summary
roles
background 3polarities
background 3representative citing papers
Uniform diffusion models rely on a leave-one-out denoiser rather than the usual denoising posterior, with exact conversions derived; an absorbing-state reformulation is introduced that matches or exceeds masked diffusion on language modeling while preserving the original joint distribution.
Equivalence between Gaussian processes and linear diffusion models enables general conditioning on arbitrary pointwise likelihoods via ODE dynamics and Monte Carlo guidance approximation.
FSF-DMD replaces the fake-score network in distribution matching distillation with a generator-induced pseudo-velocity surrogate for flow-map generators, showing improved FID on ImageNet-1K 256x256.
Metropolis-adjusted Langevin correctors using score-based acceptance probabilities, including an exact Bernoulli factory method and a Simpson's rule approximation, reduce sampling bias in diffusion models and improve FID scores.
DiLO turns diffusion sampling into deterministic latent optimization to satisfy the manifold consistency requirement for neural operators in inverse problem solving.
Derives exact guidance transition rates for discrete flow matching models that require only one model evaluation per sampling step and unify prior approximation-based methods.
PG-DLM applies particle Gibbs sampling over full trajectories in diffusion language models to enable iterative refinement, yielding higher accuracy on reward-guided generation with theoretical convergence guarantees.
Reformulates constrained black-box optimization as posterior inference in latent space of flow-based models amortized by outsourced diffusion models, claiming superior performance on synthetic and real tasks.
In the Gaussian setting the Wasserstein error of score-matching-plus-diffusion sampling equals a kernel norm of the data power spectrum whose kernel is determined by the four error sources and the algorithm parameters.
SOM is an actor-critic algorithm that constructs the target velocity field for one-step MeanFlow policies directly from the Q-function via score estimation and probability flow ODE, achieving claimed SOTA on locomotion tasks with reduced training and inference time.
GOAL uses conditioned diffusion on relational graphs with typed edges to produce feasible multi-objective solutions for scheduling problems, reporting 100% feasibility and sub-0.2% MAPE on FSP, JSP, and FJSP up to 20 jobs.
PGID restores watermark detection in diffusion models by using progressive inversion-denoising cycles to correct latents displaced by removal or forgery attacks.
CrystalREPA closes the representation gap between crystal generators and universal MLIPs via contrastive alignment, yielding more stable and valid generated crystals while revealing that MLIP teacher quality is better predicted by representation distinguishability than by leaderboard accuracy.
SymDrift makes drifting models produce symmetry-invariant samples in one step via symmetrized coordinate drifts or G-invariant embeddings, outperforming prior one-shot baselines on molecular benchmarks and cutting compute by up to 40x.
A variational method learns a neural approximation to the conditional backward-in-time score of the posterior SDE, inducing an ELBO for joint smoothing and parameter learning from sparse data.
The Transformer is recovered exactly as the forward Euler step of spherical SVFlow, with multi-head attention and MoE/FFN as approximations to its vector field.
Discrete decentralized learning dynamics on manifolds converge uniformly to an overdamped Langevin SDE whose stationary states produce orthogonally disentangled, linearly separable features.
NPN introduces a neural-network-based regularization that promotes reconstructions lying in a low-dimensional projection of the sensing operator's null-space, with claimed theoretical guarantees and improved empirical performance across compressive sensing, deblurring, super-resolution, CT, and MRI.
Derives closed-form optimal loss for unified diffusion models, provides variance-controlled estimators, and shows improved diagnosis, training schedules, and power-law scaling after subtracting the optimal value.
Latent Wavelet Diffusion uses wavelet energy map masking and a scale-consistent VAE to improve detail fidelity in 2K-4K image generation without extra inference overhead.
CT-OT Flow estimates continuous-time dynamics from discrete temporal snapshots by using partial optimal transport to align intervals and kernel smoothing to reconstruct distributions for ODE/SDE training.
Diffusion-based refinement followed by consistency distillation improves music source separation quality and inference speed across U-Net and BS-RoFormer backbones on Slakh2100 and MUSDB18.
SSLS combines score-based Langevin Monte Carlo with annealing for nonlinear posterior updates in sequential assimilation, supported by total-variation convergence bounds that establish asymptotic stability and numerical tests in high-dimensional nonlinear settings.
citing papers explorer
-
Uniform Diffusion Models Revisited: Leave-One-Out Denoiser and Absorbing State Reformulation
Uniform diffusion models rely on a leave-one-out denoiser rather than the usual denoising posterior, with exact conversions derived; an absorbing-state reformulation is introduced that matches or exceeds masked diffusion on language modeling while preserving the original joint distribution.
-
Discrete Guidance Matching: Exact Guidance for Discrete Flow Matching
Derives exact guidance transition rates for discrete flow matching models that require only one model evaluation per sampling step and unify prior approximation-based methods.
-
Inference-Time Scaling of Diffusion Language Models via Trajectory Refinement
PG-DLM applies particle Gibbs sampling over full trajectories in diffusion language models to enable iterative refinement, yielding higher accuracy on reward-guided generation with theoretical convergence guarantees.
-
Posterior Inference in Latent Space for Scalable Constrained Black-box Optimization
Reformulates constrained black-box optimization as posterior inference in latent space of flow-based models amortized by outsourced diffusion models, claiming superior performance on synthetic and real tasks.
-
From Score Matching to Diffusion: A Fine-Grained Error Analysis in the Gaussian Setting
In the Gaussian setting the Wasserstein error of score-matching-plus-diffusion sampling equals a kernel norm of the data power spectrum whose kernel is determined by the four error sources and the algorithm parameters.
-
Score-Based One-step MeanFlow Policy Optimization
SOM is an actor-critic algorithm that constructs the target velocity field for one-step MeanFlow policies directly from the Q-function via score estimation and probability flow ODE, achieving claimed SOTA on locomotion tasks with reduced training and inference time.
-
SymDrift: One-Shot Generative Modeling under Symmetries
SymDrift makes drifting models produce symmetry-invariant samples in one step via symmetrized coordinate drifts or G-invariant embeddings, outperforming prior one-shot baselines on molecular benchmarks and cutting compute by up to 40x.
-
Transformer as an Euler Discretization of Score-based Variational Flow
The Transformer is recovered exactly as the forward Euler step of spherical SVFlow, with multi-head attention and MoE/FFN as approximations to its vector field.
-
Continuous Limits of Coupled Flows in Representation Learning
Discrete decentralized learning dynamics on manifolds converge uniformly to an overdamped Langevin SDE whose stationary states produce orthogonally disentangled, linearly separable features.
-
Diagnosing and Improving Diffusion Models by Estimating the Optimal Loss Value
Derives closed-form optimal loss for unified diffusion models, provides variance-controlled estimators, and shows improved diagnosis, training schedules, and power-law scaling after subtracting the optimal value.
-
CT-OT Flow: Estimating Continuous-Time Dynamics from Discrete Temporal Snapshots
CT-OT Flow estimates continuous-time dynamics from discrete temporal snapshots by using partial optimal transport to align intervals and kernel smoothing to reconstruct distributions for ODE/SDE training.
-
EventFlow: Forecasting Temporal Point Processes with Flow Matching
EventFlow applies flow matching to learn joint distributions over event times for temporal point processes, reporting 20-53% lower forecast error than autoregressive baselines on standard TPP benchmarks with fewer sampling calls.