AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Synthesizable Molecular Generation via Soft-constrained GFlowNets with Rich Chemical Priors

Hyeonah Kim, Minsu Kim, Celine Roget, Dionessa Biton, Louis Vaillancourt, Yves V. Brun, Yoshua Bengio, Alex Hernández-García

ICML 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Structure-based 3D molecule generation · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add contrastive auxiliary loss on replay buffers (vs reward shaping) (ablation)RTB post-training of GP-MolFormer · sEH proxyPositive Top-100 sEH: 1.040 -> 1.043Improved
replace reward shaping with auxiliary loss (ablation)RTB post-training · PMO GSK3βAUC Top-10: 0.502 -> 0.807Improved
add genetic exploration to S3-GFN (ablation)S3-GFN · PMO GSK3βAUC Top-10: 0.807 -> 0.905Improved
add genetic exploration to reward shaping (ablation)RTB + RS · PMO GSK3βAUC Top-10: 0.502 -> 0.862Improved
remove auxiliary loss (positive-only training) (ablation)S3-GFN · 2D grid worldprobability mass on infeasible regionHurt
use SMILES-based S3-GFN instead of reaction-based GFlowNetreaction-based GFlowNets · LIT-PCBA ADRB2Vina score (kcal/mol): -11.45 -> -12.32Improved
use S3-GFN instead of reaction-based RxnFlowRxnFlow · LIT-PCBA ADRB2AiZynthFinder success rate (%): 60.25 -> 100.0Improved
use S3-GFN vs reward shaping under changed constraints (ablation)RTB + RS · sEH changed constraintsNum. unique: 770.3 -> 929.0Improved
reward shaping vs S3-GFN positive ratio under changed constraints (ablation)S3-GFN · sEH changed constraintsPositive ratio: 0.909 -> 0.883Hurt
use reward shaping with policy-gradient REINVENTREINVENT · PMO GSK3βAUC Top-10: 0.830Improved
pretrained SMILES prior with GFlowNet post-trainingGP-MolFormer prior · sEH proxyPositive ratio: 0.299 -> 0.945Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy