Data, Results, and Configurations for "ResT-LAVA: Data Valuation for Treatment Effects" (ACML 2026 Submission)

For data transparency and reproducibility, this repository provides the datasets, experimental results, and model configurations for the paper "ResT-LAVA: Data Valuation for Treatment Effects," submitted to ACML 2026. This includes: Processed Datasets: Includes synthetic and semi-synthetic datasets (OpenML: Pol, Fried, 2dplanes; RealCause: IHDP, LaLonde-CPS) processed using our causal data-generating process. Corruption-injected variants are provided to facilitate reproducibility of our robustness experiments. Experimental Results: Comprehensive result CSVs containing granular data beyond the summary tables presented in the paper. Reproducibility Assets: Includes model checkpoints and config files used for experiments, structured to facilitate easier re-runs of the paper’s primary results.