MolMem achieves 90% success on single-property molecular optimization tasks (1.5x over best baseline) and 52% on multi-property tasks with only 500 oracle calls via static exemplar memory and evolving skill memory in agentic RL.
[Action] [What] [Where (if clear)] to [Effect]
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MolMem: Memory-Augmented Agentic Reinforcement Learning for Sample-Efficient Molecular Optimization
MolMem achieves 90% success on single-property molecular optimization tasks (1.5x over best baseline) and 52% on multi-property tasks with only 500 oracle calls via static exemplar memory and evolving skill memory in agentic RL.