Temporal fine-tuning for early risk detection
Horacio Thompson, Esaú Villatoro-Tello, Manuel Montes-y-Gómez, Marcelo Errecalde · 2025
Read Temporal fine-tuning for early risk detection by Horacio Thompson, Esaú Villatoro-Tello, Manuel Montes-y-Gómez, Marcelo Errecalde (2025) 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-2505-11280.txt.
Description
Horacio Thompson and co-authors wrote a 2025 paper on spotting online risk early. The task is to flag users who may face health or social harm, post by post. A good system must be both right and fast, since a late flag can miss the chance to help.
The authors propose a new way to train transformer models that builds time into the training itself. The method reads a full post history and judges the model with scores that count both right calls and delay. Tests used Spanish posts tied to low mood and eating disorders. The method matched the best systems from a 2023 shared task on those topics.
Catalog record
| CREATOR | Horacio Thompson, Esaú Villatoro-Tello, Manuel Montes-y-Gómez, Marcelo Errecalde |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-11280