Research map / Probabilistic and scientific modeling / Molecules and proteins
Flow Network based Generative Models for Non-Iterative Diverse Candidate Generation
Emmanuel Bengio, Moksh Jain, Maksym Korablyov, Doina Precup, Yoshua Bengio
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use GFlowNet instead of MARS | Molecule discovery · 1800 docking evaluations | top-10 reward: 8.27 ± 0.20 -> 8.83 ± 0.15 | Improved |
| Use GFlowNet instead of MARS | Molecule discovery · 1800 docking evaluations | top-100 reward: 7.08 ± 0.13 -> 7.76 ± 0.11 | Improved |
| Use GFlowNet instead of PPO or MARS | Molecule discovery · top 1000 samples | mean pairwise Tanimoto similarity: 0.62±0.03; 0.59 ± 0.02 -> 0.44± 0.01 | Improved |
| Use GFlowNet instead of MARS | Multi-round molecule discovery · final round | mean pairwise Tanimoto similarity: 0.64 ± 0.03 -> 0.54 ± 0.04 | Improved |
| Use GFlowNet rather than MCMC | 4-D hyper-grid · multi-round mode discovery | rounds to discover 10 modes: 10 rounds -> 5 rounds | Improved |
| Use GFlowNet rather than PPO | 4-D hyper-grid · multi-round mode discovery | modes discovered: 8 -> 10 | Improved |
| Increase the terminal-transition loss weight (ablation) | GFlowNet · molecule domain | training stability | Improved |
| Raise PPO entropy regularization to discover modes (ablation) | PPO · 4-D hyper-grid | mode discovery: 0.5 | Mixed |
| Increase PPO reward β above 4 (ablation) | PPO · molecule domain | performance: 4 | No clear effect |
Closest papers by meaning
- Generative Flows on Synthetic Pathway for Drug Design (ICLR 2025 · similarity 0.90)
- Pre-Training and Fine-Tuning Generative Flow Networks (ICLR 2024 · similarity 0.90)
- Generative Flow Networks for Lead Optimization in Drug Design (Student Abstract) (AAAI 2025 · similarity 0.90)
- Flow Density Control: Generative Optimization Beyond Entropy-Regularized Fine-Tuning (NeurIPS 2025 · similarity 0.90)
- Pretraining Generative Flow Networks with Inexpensive Rewards for Molecular Graph Generation (ICML 2025 · similarity 0.90)
- Flow Factorization for Efficient Generative Flow Networks (AAAI 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
