Understanding multi-fidelity training of machine-learned force-fields
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866918424166268928 |
|---|---|
| author | Gardner, John L. A. Schulz, Hannes Helie, Jean Sun, Lixin Simm, Gregor N. C. |
| author_facet | Gardner, John L. A. Schulz, Hannes Helie, Jean Sun, Lixin Simm, Gregor N. C. |
| contents | This study systematically investigates two multi-fidelity strategies used to train machine-learned force fields (MLFFs) -- pre-training/fine-tuning and multi-headed training -- and elucidates the mechanisms underpinning their success. For pre-training and fine-tuning, we uncover a log-log linear relationship between pre-trained and fine-tuned accuracies that holds across model architectures, model sizes, and quantum-chemical methods. The success of this approach hinges on the quantity and quality of available pre-training data, and, critically, the inclusion of force labels. We demonstrate that pre-trained representations are inherently method-specific, requiring adaptation of the model backbone during fine-tuning. In contrast, multi-headed models learn method-independent backbone representations, where again the heads' accuracies are log-log linearly related. Relative to pre-training and fine-tuning, these shared representations marginally reduce model performance in most cases. However, this trade-off is offset by practical advantages: multi-headed training extends naturally to multiple labelling methods and enables partial replacement of expensive labels with cheaper alternatives, paving the way towards cost-efficient universal MLFFs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_14963 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Understanding multi-fidelity training of machine-learned force-fields Gardner, John L. A. Schulz, Hannes Helie, Jean Sun, Lixin Simm, Gregor N. C. Chemical Physics Computational Physics This study systematically investigates two multi-fidelity strategies used to train machine-learned force fields (MLFFs) -- pre-training/fine-tuning and multi-headed training -- and elucidates the mechanisms underpinning their success. For pre-training and fine-tuning, we uncover a log-log linear relationship between pre-trained and fine-tuned accuracies that holds across model architectures, model sizes, and quantum-chemical methods. The success of this approach hinges on the quantity and quality of available pre-training data, and, critically, the inclusion of force labels. We demonstrate that pre-trained representations are inherently method-specific, requiring adaptation of the model backbone during fine-tuning. In contrast, multi-headed models learn method-independent backbone representations, where again the heads' accuracies are log-log linearly related. Relative to pre-training and fine-tuning, these shared representations marginally reduce model performance in most cases. However, this trade-off is offset by practical advantages: multi-headed training extends naturally to multiple labelling methods and enables partial replacement of expensive labels with cheaper alternatives, paving the way towards cost-efficient universal MLFFs. |
| title | Understanding multi-fidelity training of machine-learned force-fields |
| topic | Chemical Physics Computational Physics |
| url | https://arxiv.org/abs/2506.14963 |