mlsanityData to 7 Oct 2026

Towards Understanding and Improving GFlowNet Training

arXiv 2305.07170 ↗First posted 11 May 2023cs.LGcited by 74

Not on the map. Why?

Abstract (arXiv)

Generative flow networks (GFlowNets) are a family of algorithms that learn a generative policy to sample discrete objects x with non-negative reward R(x). Learning objectives guarantee the GFlowNet samples x from the target distribution p^*(x) propto R(x) when loss is globally minimized over all states or trajectories, but it is unclear how well they perform with practical limits on training resources. We introduce an efficient evaluation strategy to compare the learned sampling distribution to the target reward distribution. As flows can be underdetermined given training data, we clarify the importance of learned flows to generalization and matching p^*(x) in practice. We investigate how to learn better flows, and propose (i) prioritized replay training of high-reward x, (ii) relative edge flow policy parametrization, and (iii) a novel guided trajectory balance objective, and show how it can solve a substructure credit assignment problem. We substantially improve sample efficiency on biochemical design tasks.

Key facts. Cited by 74 papers in the map (data through 7 Oct 2026). Builds on 12 of its 18 references that are in the map. First posted on arXiv 11 May 2023.

Builds on

12 of 18 references.

References · 12

  1. 2015Prioritized Experience Replaycited by 1,070
  2. 2021Flow Network based Generative Models for Non-Iterative Diverse Candidate Generationcited by 384
  3. 2021GFlowNet Foundationscited by 286
  4. 2022Trajectory balance: Improved credit assignment in GFlowNetscited by 244
  5. 2022Biological Sequence Design with GFlowNetscited by 172
  6. 2020Revisiting Fundamentals of Experience Replaycited by 171
  7. 2022Bayesian Structure Learning with Generative Flow Networkscited by 153
  8. 2022Learning GFlowNets from partial episodes for improved convergence and stabilitycited by 129
  9. 2022Generative Flow Networks for Discrete Probabilistic Modelingcited by 113
  10. 2022Design-Bench: Benchmarks for Data-Driven Offline Model-Based Optimizationcited by 96
  11. 2019Measuring Compositionality in Representation Learningcited by 84
  12. 2022Improving Generative Flow Networks with Path Regularizationcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 14; 2024: 28; 2025: 14; 2026: 18.

74 papers.

Sort:

By area:

  1. 2023Protein Design with Guided Discrete Diffusioncited by 154
  2. 2023Amortizing intractable inference in large language modelscited by 94
  3. 2023Generative Flow Networks as Entropy-Regularized RLcited by 60
  4. 2023Diffusion Generative Flow Samplers: Improving learning signals through partial trajectory optimizationcited by 57
  5. 2024Improved off-policy training of diffusion samplerscited by 47
  6. 2023Local Search GFlowNetscited by 45
  7. 2023Let the Flows Tell: Solving Graph Combinatorial Optimization Problems with GFlowNetscited by 40
  8. 2024Discrete Probabilistic Inference as Control in Multi-path Environmentscited by 40
  9. 2025FlowRL: Matching Reward Distributions for LLM Reasoningcited by 38
  10. 2024Genetic-guided GFlowNets for Sample Efficient Molecular Optimizationcited by 36
  11. 2023Maximum entropy GFlowNets with soft Q-learningcited by 30
  12. 2024Efficient Diversity-Preserving Diffusion Alignment via Gradient-Informed GFlowNetscited by 30
38 more
  1. 2024Improving GFlowNets for Text-to-Image Diffusion Alignmentcited by 27
  2. 2025Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Trainingcited by 27
  3. 2023Learning to Scale Logits for Temperature-Conditional GFlowNetscited by 26
  4. 2023Learning Energy Decompositions for Partial Inference of GFlowNetscited by 24
  5. 2024Adaptive teachers for amortized samplerscited by 23
  6. 2023An Empirical Study of the Effectiveness of Using a Replay Buffer on Mode Discovery in GFlowNetscited by 17
  7. 2025Learning to Sample Effective and Diverse Prompts for Text-to-Image Generationcited by 16
  8. 2024PROUD: PaRetO-gUided Diffusion Model for Multi-objective Generationcited by 15
  9. 2024Sampling from Energy-based Policies using Diffusioncited by 15
  10. 2023Order-Preserving GFlowNetscited by 13
  11. 2024Pessimistic Backward Policy for GFlowNetscited by 13
  12. 2024Improved Off-policy Reinforcement Learning in Biological Sequence Designcited by 13
  13. 2024Optimizing Backward Policies in GFlowNets via Trajectory Likelihood Maximizationcited by 12
  14. 2024RetroGFN: Diverse and Feasible Retrosynthesis using GFlowNetscited by 10
  15. 2025Adversarial Generative Flow Network for Solving Vehicle Routing Problemscited by 10
  16. 2024Beyond Squared Error: Exploring Loss Design for Enhanced Training of Generative Flow Networkscited by 9
  17. 2023Expert-Aided Causal Discovery of Ancestral Graphscited by 8
  18. 2024Looking Backward: Retrospective Backward Synthesis for Goal-Conditioned GFlowNetscited by 8
  19. 2025Revisiting Non-Acyclic GFlowNets in Discrete Environmentscited by 8
  20. 2025Can Knowledge-Graph-based Retrieval Augmented Generation Really Retrieve What You Need?cited by 8
  21. 2023DGFN: Double Generative Flow Networkscited by 7
  22. 2023Symmetric Replay Training: Enhancing Sample Efficiency in Deep Reinforcement Learning for Combinatorial Optimizationcited by 6
  23. 2024Investigating Generalization Behaviours of Generative Flow Networkscited by 6
  24. 2024MetaGFN: Exploring Distant Modes with Adapted Metadynamics for Continuous GFlowNetscited by 6
  25. 2024On Divergence Measures for Training GFlowNetscited by 6
  26. 2024GDPO: Learning to Directly Align Language Models with Diversity Using GFlowNetscited by 6
  27. 2024Random Policy Evaluation Uncovers Policies of Generative Flow Networkscited by 5
  28. 2025Scalable and Cost-Efficient de Novo Template-Based Molecular Generationcited by 5
  29. 2025Reinforced sequential Monte Carlo for amortised samplingcited by 5
  30. 2024GFlowNet Training by Policy Gradientscited by 4
  31. 2024Streaming Bayes GFlowNetscited by 4
  32. 2025Symmetry-Aware GFlowNetscited by 4
  33. 2024Evolution Guided Generative Flow Networkscited by 3
  34. 2024Embarrassingly Parallel GFlowNetscited by 3
  35. 2025Deep Symbolic Optimization: Reinforcement Learning for Symbolic Mathematicscited by 3
  36. 2026Avoid What You Know: Divergent Trajectory Balance for GFlowNetscited by 3
  37. 2025gfnx: Fast and Scalable Library for Generative Flow Networks in JAXcited by 2
  38. 2026Discrete diffusion samplers and bridges: Off-policy algorithms and applications in latent spacescited by 2

50 most cited of 74 All 74 (JSON)