A Causal Foundation Model for Structure and Outcome Prediction
Max Zhu, Martino Mansoldo, Ching-Hao Wang, Stefan Groha · 2026
Read A Causal Foundation Model for Structure and Outcome Prediction by Max Zhu, Martino Mansoldo, Ching-Hao Wang, Stefan Groha (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-2606-26467.txt.
Description
Max Zhu, Martino Mansoldo, Ching-Hao Wang, and Stefan Groha introduce TabPFN-CFM, a causal foundation model. A foundation model is one large pre-trained system reused for many tasks. Causal models go beyond correlation. They predict what happens when someone intervenes in a system. From observational data alone, it predicts causal structure and outcomes alike.
The model answers questions at all three levels of Pearl's causal hierarchy. Those levels cover seeing, doing, and imagining what would have happened. It can also use a known causal graph when one is available. A causal graph maps which variables drive which others. Training used synthetic data sets, yet the model generalised to real ones. It beat baselines at both structure prediction and outcome prediction. A refined training procedure also cut training cost almost fourfold, the authors report.
Catalog record
| CREATOR | Max Zhu, Martino Mansoldo, Ching-Hao Wang, Stefan Groha |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.26467