The paper introduces a new learning objective called trajectory balance for Generative Flow Networks (GFlowNets), aiming to improve credit assignment across long action sequences. It demonstrates that minimizing this objective yields a policy that samples exactly from the target distribution. Experiments on four domains show that trajectory balance enhances convergence, sample diversity, and robustness to long sequences and large action spaces.
By Esmeralda S. Whitammer, Moksh Jain, Emmanuel Bengio, Chen Sun, Yoshua Bengio
arXiv:2608. 03967v1 Announce Type: cross Abstract: Generative Flow Networks (GFlowNets) have emerged as a flexible framework for amortised inference over discrete and mixed discrete-continuous objects, requiring only an unnormalised target density specified through a reward.
By Yordan Raykov, Rodrigo Veiga
arXiv:2605. 01729v2 Announce Type: replace Abstract: Generative Flow Networks (GFlowNets) learn to sample states proportional to an unnormalized reward.
By Zengxiang Lei, Ananth Shreekumar, Jonathan Rosenthal, Ruoyu Song, Alvaro A. Cardenas, Daniel J. Fremont, Dongyan Xu, Satish Ukkusuri, Z. Berkay Celik
The paper introduces Particle GFlowNets, showing that Generative Marginalization Models (MaMs) are equivalent to Generative Flow Networks. It extends MaMs to non‑autoregressive sampling and proposes an automatic full‑state rejuvenation criterion based on the Gelman‑Rubin statistic to accelerate learning. Experiments demonstrate significant training speedups in large combinatorial spaces.
By Tiago da Silva, Diego Mesquita, Salem Lahlou
arXiv:2606. 06272v1 Announce Type: new Abstract: Generative Flow Networks (GFlowNets) are a framework for sampling structured objects via stochastic trajectories in a directed graph.
By Ian Maksimov, Nikita Morozov, Denis Belomestny, Sergey Samsonov
arXiv:2607. 00535v1 Announce Type: cross Abstract: Few-step flow-map generators, such as consistency models and MeanFlow, accelerate sampling by directly learning long-range transport maps between noise and data.
By Zhiqi Li, Wen Zhang, Bo Zhu
The paper introduces Flow Divergence Sampler (FDS), a training‑free method that refines intermediate states in flow‑matching models by using the divergence of the marginal velocity field to detect and correct misguidance toward low‑density regions. FDS operates during inference, requires no additional training, and can be applied as a plug‑and‑play module with standard solvers and existing flow backbones. Experiments show that FDS consistently improves fidelity in tasks such as text‑to‑image synthesis and inverse problems.
By Yeonwoo Cha, Jaehoon Yoo, Semin Kim, Yunseo Park, Jinhyeon Kwon, Seunghoon Hong
arXiv:2603. 10718v3 Announce Type: replace Abstract: Flow Matching enables simulation-free training of generative models on Riemannian manifolds, yet sampling typically still relies on numerically integrating a probability-flow ODE.
By Zichen Zhong, Haoliang Sun, Yukun Zhao, Yongshun Gong, Yilong Yin
arXiv:2606. 15793v1 Announce Type: cross Abstract: This paper explores policy gradient algorithms for training stochastic policies to sample from structured discrete probability distributions under the Generative Flow Network (GFlowNet) framework.
By Anna Zykova-Myzina, Timofei Gritsaev, Daniil Tiapkin, Nikita Morozov
The paper introduces Bayesian Flow Networks for Offline Trajectory Planning (BFN-RL), a generative modeling framework that unifies discrete and continuous trajectory synthesis for offline reinforcement learning. Unlike prior diffusion models that rely on Gaussian noise, BFN-RL iteratively updates distribution parameters, enabling a categorical planner to produce future state sequences and an inverse-dynamics model to translate these states into actions. Experiments demonstrate that BFN-RL effectively generates trajectories in both discrete planning and continuous control tasks, highlighting its versatility across data modalities.
By Ludvig Killingberg, Helge Langseth
The paper investigates training diffusion models to sample from distributions defined by unnormalized densities or energy functions. It benchmarks various diffusion-structured inference techniques, including simulation-based variational methods and off-policy approaches such as continuous generative flow networks, highlighting their relative strengths and challenging some prior claims. Additionally, the authors introduce a new exploration strategy for off-policy methods that employs local search in the target space with a replay buffer, demonstrating improved sample quality across multiple target distributions.
By Marcin Sendera, Minsu Kim, Sarthak Mittal, Pablo Lemos, Luca Scimeca, Jarrid Rector-Brooks, Alexandre Adam, Yoshua Bengio, Esmeralda S. Whitammer
In rectified-flow-based generative models, the neural network can be trained to predict two different targets, such as the instantaneous velocity or the data endpoint, to perform denoising. Although prior work shows that these parameterizations lead to different empirical behaviors, the mechanisms underlying their respective advantages remain to be underexplored, and how to combine them effectively is still unclear.