A Study on the Fine-Tuning Performance of Universal Machine-Learned Interatomic Potentials (U-MLIPs)
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913885064265728 |
|---|---|
| author | Liu, Xiaoqing Zeng, Kehan Wang, Yangshuai Zhao, Teng |
| author_facet | Liu, Xiaoqing Zeng, Kehan Wang, Yangshuai Zhao, Teng |
| contents | Universal machine-learned interatomic potentials (U-MLIPs) have demonstrated effectiveness across diverse atomistic systems but often require fine-tuning for task-specific accuracy. We investigate the fine-tuning of two MACE-based foundation models, MACE-MP-0 and its variant MACE-MP-0b, and identify key insights. Fine-tuning on task-specific datasets enhances accuracy and, in some cases, outperforms models trained from scratch. Additionally, fine-tuned models benefit from faster convergence due to the strong initial predictions provided by the foundation model. The success of fine-tuning also depends on careful dataset selection, which can be optimized through filtering or active learning. We further discuss practical strategies for achieving better fine-tuning foundation models in atomistic simulations and explore future directions for their development and applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_07401 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A Study on the Fine-Tuning Performance of Universal Machine-Learned Interatomic Potentials (U-MLIPs) Liu, Xiaoqing Zeng, Kehan Wang, Yangshuai Zhao, Teng Computational Physics Universal machine-learned interatomic potentials (U-MLIPs) have demonstrated effectiveness across diverse atomistic systems but often require fine-tuning for task-specific accuracy. We investigate the fine-tuning of two MACE-based foundation models, MACE-MP-0 and its variant MACE-MP-0b, and identify key insights. Fine-tuning on task-specific datasets enhances accuracy and, in some cases, outperforms models trained from scratch. Additionally, fine-tuned models benefit from faster convergence due to the strong initial predictions provided by the foundation model. The success of fine-tuning also depends on careful dataset selection, which can be optimized through filtering or active learning. We further discuss practical strategies for achieving better fine-tuning foundation models in atomistic simulations and explore future directions for their development and applications. |
| title | A Study on the Fine-Tuning Performance of Universal Machine-Learned Interatomic Potentials (U-MLIPs) |
| topic | Computational Physics |
| url | https://arxiv.org/abs/2506.07401 |