Causal Foundation Models: Pretrained Nets Estimate Treatment Effects via In-Context Learning

chaumian · x · 2026-09-04

An arXiv paper introduces Causal Foundation Models (CFMs), bringing the foundation-model paradigm to causal inference. Instead of bespoke pipelines per problem, CFMs are pretrained neural networks that estimate causal quantities like average treatment effects on entirely new datasets via in-context learning, with no model updates. The paper is a practical introduction with background on causal inference and ML, plus example code and Jupyter notebooks.

Original post →

More from Research

Research channel →