Predictive Feature Caching for Training-free Acceleration of Molecular Geometry Generation
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918154842669056 |
|---|---|
| author | Sommer, Johanna Rachwan, John Fleischmann, Nils Günnemann, Stephan Charpentier, Bertrand |
| author_facet | Sommer, Johanna Rachwan, John Fleischmann, Nils Günnemann, Stephan Charpentier, Bertrand |
| contents | Flow matching models generate high-fidelity molecular geometries but incur significant computational costs during inference, requiring hundreds of network evaluations. This inference overhead becomes the primary bottleneck when such models are employed in practice to sample large numbers of molecular candidates. This work discusses a training-free caching strategy that accelerates molecular geometry generation by predicting intermediate hidden states across solver steps. The proposed method operates directly on the SE(3)-equivariant backbone, is compatible with pretrained models, and is orthogonal to existing training-based accelerations and system-level optimizations. Experiments on the GEOM-Drugs dataset demonstrate that caching achieves a twofold reduction in wall-clock inference time at matched sample quality and a speedup of up to 3x compared to the base model with minimal sample quality degradation. Because these gains compound with other optimizations, applying caching alongside other general, lossless optimizations yield as much as a 7x speedup. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_04646 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Predictive Feature Caching for Training-free Acceleration of Molecular Geometry Generation Sommer, Johanna Rachwan, John Fleischmann, Nils Günnemann, Stephan Charpentier, Bertrand Machine Learning Artificial Intelligence Flow matching models generate high-fidelity molecular geometries but incur significant computational costs during inference, requiring hundreds of network evaluations. This inference overhead becomes the primary bottleneck when such models are employed in practice to sample large numbers of molecular candidates. This work discusses a training-free caching strategy that accelerates molecular geometry generation by predicting intermediate hidden states across solver steps. The proposed method operates directly on the SE(3)-equivariant backbone, is compatible with pretrained models, and is orthogonal to existing training-based accelerations and system-level optimizations. Experiments on the GEOM-Drugs dataset demonstrate that caching achieves a twofold reduction in wall-clock inference time at matched sample quality and a speedup of up to 3x compared to the base model with minimal sample quality degradation. Because these gains compound with other optimizations, applying caching alongside other general, lossless optimizations yield as much as a 7x speedup. |
| title | Predictive Feature Caching for Training-free Acceleration of Molecular Geometry Generation |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2510.04646 |