Mol-MoE: Training Preference-Guided Routers for Molecule Generation

TL;DR

Mol-MoE introduces a mixture-of-experts architecture with preference-guided routers, enabling flexible, test-time steering of molecule generation to optimize multiple properties without retraining, thus improving efficiency and trade-off exploration.

Contribution

The paper presents Mol-MoE, a novel MoE-based model with preference-guided routing that allows for efficient, multi-objective molecule generation without retraining.

Findings

Achieves superior sample quality compared to state-of-the-art methods.

Enables rapid test-time trade-off exploration of molecular properties.

Demonstrates effective multi-objective optimization without retraining.

Abstract

Recent advances in language models have enabled framing molecule generation as sequence modeling. However, existing approaches often rely on single-objective reinforcement learning, limiting their applicability to real-world drug design, where multiple competing properties must be optimized. Traditional multi-objective reinforcement learning (MORL) methods require costly retraining for each new objective combination, making rapid exploration of trade-offs impractical. To overcome these limitations, we introduce Mol-MoE, a mixture-of-experts (MoE) architecture that enables efficient test-time steering of molecule generation without retraining. Central to our approach is a preference-based router training objective that incentivizes the router to combine experts in a way that aligns with user-specified trade-offs. This provides improved flexibility in exploring the chemical property space at test time, facilitating rapid trade-off exploration. Benchmarking against state-of-the-art methods, we show that Mol-MoE achieves superior sample quality and steerability.