🔗 CDFM: Causal Discovery Foundation Model
Zero-shot causal discovery from observational tabular data. Upload a CSV file and CDFM predicts the causal graph in a single forward pass — no training required.
Examples
How it works: CDFM is a ~9.6M parameter foundation model pretrained on a massive, diverse space of synthetic structural causal models. Given observational data X (N × D), it predicts the causal graph G (D × D) in a single forward pass.
The model handles mixed numeric/categorical data and missing values automatically. Upload any CSV where rows are samples and columns are variables.