Large Causal Models for Temporal Causal Discovery

TL;DR

This paper introduces large causal models (LCMs) designed for temporal causal discovery, enabling scalable, accurate, and generalizable causal inference across multiple datasets with deep neural architectures.

Contribution

The paper presents a novel framework for LCMs that combines synthetic and real data, scaling causal discovery to larger variable counts and architectures, outperforming prior small-scale approaches.

Findings

LCMs scale effectively to higher variable counts

LCMs maintain strong performance on diverse benchmarks

Trained models outperform classical and neural baselines

Abstract

Causal discovery for both cross-sectional and temporal data has traditionally followed a dataset-specific paradigm, where a new model is fitted for each individual dataset. Such an approach limits the potential of multi-dataset pretraining. The concept of large causal models (LCMs) envisions a class of pre-trained neural architectures specifically designed for temporal causal discovery. Prior approaches are constrained to small variable counts, degrade with larger inputs, and rely heavily on synthetic data, limiting generalization. We propose a principled framework for LCMs, combining diverse synthetic generators with realistic time-series datasets, allowing learning at scale. Extensive experiments on synthetic, semi-synthetic and realistic benchmarks show that LCMs scale effectively to higher variable counts and deeper architectures while maintaining strong performance. Trained models achieve competitive or superior accuracy compared to classical and neural baselines, particularly in out-of-distribution settings, while enabling fast, single-pass inference. Results demonstrate LCMs as a promising foundation-model paradigm for temporal causal discovery. Experiments and model weights are available at https://github.com/kougioulis/LCM-paper/.