TriForces: Augmenting Atomistic GNNs for Transferable Representations
Ali Ramlaoui, Alexandre Duval, Hannah Bull, Victor Schmidt, Hugues Talbot, Fragkiskos D. Malliaros, Joseph Musielewicz · 2026
Read TriForces: Augmenting Atomistic GNNs for Transferable Representations by Ali Ramlaoui, Alexandre Duval, Hannah Bull, Victor Schmidt, Hugues Talbot, Fragkiskos D. Malliaros, Joseph Musielewicz (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-20581.txt.
Description
Ali Ramlaoui and six colleagues present TriForces, a three-stream framework to help atom models transfer better across new chemistries. Learned potentials for atoms do well with large training sets but adapt unevenly to small, costly new sets. TriForces keeps composition facts apart from structure facts and uses self-taught learning to keep traits that transfer. On one materials test, it cut energy error by 57 percent with only 20,000 samples and improved force error at all sample sizes.
Catalog record
| CREATOR | Ali Ramlaoui, Alexandre Duval, Hannah Bull, Victor Schmidt, Hugues Talbot, Fragkiskos D. Malliaros, Joseph Musielewicz |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80246 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.20581