Stochastic generative flow networks

[PDF] mlr.press

Better training of gflownets with local credit and incomplete trajectories

L Pan, N Malkin, D Zhang… - … Conference on Machine …, 2023 - proceedings.mlr.press
Abstract Generative Flow Networks or GFlowNets are related to Monte-Carlo Markov chain
methods (as they sample from a distribution specified by an energy function), reinforcement …
被引用次数：49 相关文章 所有 6 个版本
[PDF] arxiv.org

Diffusion generative flow samplers: Improving learning signals through partial trajectory optimization

D Zhang, RTQ Chen, CH Liu, A Courville… - arXiv preprint arXiv …, 2023 - arxiv.org
We tackle the problem of sampling from intractable high-dimensional density functions, a
fundamental task that often appears in machine learning and statistics. We extend recent …
被引用次数：20 相关文章 所有 3 个版本
[PDF] arxiv.org

Local search gflownets

M Kim, T Yun, E Bengio, D Zhang, Y Bengio… - arXiv preprint arXiv …, 2023 - arxiv.org
Generative Flow Networks (GFlowNets) are amortized sampling methods that learn a
distribution over discrete objects proportional to their rewards. GFlowNets exhibit a …
被引用次数：20 相关文章 所有 3 个版本
[PDF] arxiv.org

Learning to scale logits for temperature-conditional gflownets

M Kim, J Ko, T Yun, D Zhang, L Pan, W Kim… - arXiv preprint arXiv …, 2023 - arxiv.org
GFlowNets are probabilistic models that sequentially generate compositional structures
through a stochastic policy. Among GFlowNets, temperature-conditional GFlowNets can …
被引用次数：13 相关文章 所有 4 个版本
[PDF] arxiv.org

Pre-training and fine-tuning generative flow networks

L Pan, M Jain, K Madan, Y Bengio - arXiv preprint arXiv:2310.03419, 2023 - arxiv.org
Generative Flow Networks (GFlowNets) are amortized samplers that learn stochastic
policies to sequentially generate compositional objects from a given unnormalized reward …
被引用次数：10 相关文章 所有 4 个版本
[PDF] mlr.press

Maximum entropy GFlowNets with soft Q-learning

S Mohammadpour, E Bengio… - International …, 2024 - proceedings.mlr.press
Abstract Generative Flow Networks (GFNs) have emerged as a powerful tool for sampling
discrete objects from unnormalized distributions, offering a scalable alternative to Markov …
被引用次数：8 相关文章 所有 3 个版本
[PDF] arxiv.org

Discrete probabilistic inference as control in multi-path environments

T Deleu, P Nouri, N Malkin, D Precup… - arXiv preprint arXiv …, 2024 - arxiv.org
We consider the problem of sampling from a discrete and structured distribution as a
sequential decision problem, where the objective is to find a stochastic policy such that …
被引用次数：15 相关文章 所有 3 个版本
[PDF] arxiv.org

Distributional gflownets with quantile flows

D Zhang, L Pan, RTQ Chen, A Courville… - arXiv preprint arXiv …, 2023 - arxiv.org
Generative Flow Networks (GFlowNets) are a new family of probabilistic samplers where an
agent learns a stochastic policy for generating complex combinatorial structure through a …
被引用次数：18 相关文章 所有 4 个版本
[PDF] arxiv.org

Expected flow networks in stochastic environments and two-player zero-sum games

M Jiralerspong, B Sun, D Vucetic, T Zhang… - arXiv preprint arXiv …, 2023 - arxiv.org
Generative flow networks (GFlowNets) are sequential sampling models trained to match a
given distribution. GFlowNets have been successfully applied to various structured object …
被引用次数：2 相关文章 所有 3 个版本
[PDF] arxiv.org

Qgfn: Controllable greediness with action values

E Lau, SZ Lu, L Pan, D Precup, E Bengio - arXiv preprint arXiv:2402.05234, 2024 - arxiv.org
Generative Flow Networks (GFlowNets; GFNs) are a family of reward/energy-based
generative methods for combinatorial objects, capable of generating diverse and high-utility …
被引用次数：6 相关文章 所有 2 个版本