Mol-MoE: Training Preference-Guided Routers for Molecule Generation
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866912225535459328 |
|---|---|
| author | Calanzone, Diego D'Oro, Pierluca Bacon, Pierre-Luc |
| author_facet | Calanzone, Diego D'Oro, Pierluca Bacon, Pierre-Luc |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_05633 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Mol-MoE: Training Preference-Guided Routers for Molecule Generation Calanzone, Diego D'Oro, Pierluca Bacon, Pierre-Luc Machine Learning 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. |
| title | Mol-MoE: Training Preference-Guided Routers for Molecule Generation |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2502.05633 |