16 papers
LongTake: Learning to Sustain Dynamics in Long-Horizon Video Generation
Byoungwoo Park, Jaemoo Choi, Juho Lee +1
World models, game simulators, and long-take video creation require coherent scene evolution and sustained dynamics over extended durations. Autoregressive (AR) video diffusion pro…
Hard-Constrained Sampling on Embedded Riemannian Manifolds via Adjoint Schrödinger Bridges
Mattia Mosso, Jaemoo Choi, Heng Yang
A variety of tasks require sampling from unnormalized Boltzmann distributions supported on manifolds. Building upon the foundations of adjoint matching and adjoint Schrödinger brid…
Scaling Reinforcement Learning for Diffusion Models via Velocity Matching
Jaemoo Choi, Wei Guo, Yuchen Zhu +4
Reward-based fine-tuning of diffusion models has largely inherited likelihood-based policy optimization developed for autoregressive large language models (LLMs). Diffusion models,…
Coarse-to-Fine Compositional Diffusion for Long-Horizon Planning
Byoungwoo Park, Utkarsh A. Mishra, Jaemoo Choi +2
Diffusion models provide strong priors for generating structured data, but many tasks require outputs beyond the scale on which these models are typically trained. Compositional ge…
RMA: Context-Orchestrated Research Math Agents
Zelin Zhao, Bo Yuan, Yuchen Zhu +2
Long-horizon mathematical reasoning fails less often because a model cannot produce a valid next step than because an agent fails to maintain and expose the right semantic state ac…
MetaDNS: Enhancing Exploration in Discrete Neural Samplers via Well-Tempered Metadynamics
Xiaochen Du, Juno Nam, Jaemoo Choi +7
Sampling from discrete distributions with multiple modes and energy barriers is fundamental to machine learning and computational physics. Recent discrete neural samplers like MDNS…
Efficient Adjoint Matching for Fine-tuning Diffusion Models
Jeongwoo Shin, Dongsoo Shin, Yuchen Zhu +5
Reward fine-tuning has become a common approach for aligning pretrained diffusion and flow models with human preferences in text-to-image generation. Among reward-gradient-based me…
Discrete Adjoint Schrödinger Bridge Sampler
Wei Guo, Yuchen Zhu, Xiaochen Du +6
Learning discrete neural samplers is challenging due to the lack of gradients and combinatorial complexity. While stochastic optimal control (SOC) and Schrödinger bridge (SB) provi…
Generalized Schrödinger Bridge on Graphs
Panagiotis Theodoropoulos, Juno Nam, Evangelos Theodorou +1
Transportation on graphs is a fundamental challenge across many domains, where decisions must respect topological and operational constraints. Despite the need for actionable polic…
Rethinking the Design Space of Reinforcement Learning for Diffusion Models: On the Importance of Likelihood Estimation Beyond Loss Design
Jaemoo Choi, Yuchen Zhu, Wei Guo +6
Reinforcement learning has been widely applied to diffusion and flow models for visual tasks such as text-to-image generation. However, these tasks remain challenging because diffu…
MFM-point: Multi-scale Flow Matching for Point Cloud Generation
Petr Molodyk, Jaemoo Choi, David W. Romero +2
In recent years, point cloud generation has gained significant attention in 3D generative modeling. Among existing approaches, point-based methods directly generate point clouds wi…
Enhancing Reasoning for Diffusion LLMs via Distribution Matching Policy Optimization
Yuchen Zhu, Wei Guo, Jaemoo Choi +4
Diffusion large language models (dLLMs) are promising alternatives to autoregressive large language models (AR-LLMs), as they potentially allow higher inference throughput. Reinfor…
Proximal Diffusion Neural Sampler
Wei Guo, Jaemoo Choi, Yuchen Zhu +2
The task of learning a diffusion-based neural sampler for drawing samples from an unnormalized target distribution can be viewed as a stochastic optimal control problem on path mea…
MDNS: Masked Diffusion Neural Sampler via Stochastic Optimal Control
Yuchen Zhu, Wei Guo, Jaemoo Choi +3
We study the problem of learning a neural sampler to generate samples from discrete state spaces where the target probability mass function is known up to…
Adjoint Schrödinger Bridge Sampler
Guan-Horng Liu, Jaemoo Choi, Yongxin Chen +2
Computational methods for learning to sample from the Boltzmann distribution -- where the target distribution is known only up to an unnormalized energy function -- have advanced s…
Non-equilibrium Annealed Adjoint Sampler
Jaemoo Choi, Yongxin Chen, Molei Tao +1
Recently, there has been significant progress in learning-based diffusion samplers, which aim to sample from a given unnormalized density. Many of these approaches formulate the sa…