Scalability of CausalMixFT Synthetic Data for Generalization on Large Cross-Domain Tabular Datasets

Fine-tuning tabular foundation models (TFMs) under data scarcity is challenging, as early stopping on even scarcer validation data often fails to capture true generalization performance. We propose CausalMixFT, a method that enhances fine-tuning robustness and downstream performance by generating structurally consistent synthetic samples using Structural Causal Models (SCMs) fitted on the target dataset. This approach augments limited real data with causally informed synthetic examples, preserving feature dependencies while expanding training diversity. Evaluated across 33 classification datas

Research goal: To what extent can CausalMixFT's synthetic data generation method scale to larger tabular datasets while maintaining consistent improvements in model generalization, as measured by metrics like AUC or F1-score on cross-domain datasets like TabM3?

Autonomous synthesis report generated by SOVEREIGN Research Kernel. Tribunal consensus score: 8.5/10.