CORTEXA
← Browse

Jiachun Li

2 papers indexed

arxivcs.LGcs.AI2026-06-30

Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation

Jiachun Li, David Simchi-Levi

Adaptive experiments for average treatment effects (ATE) require randomized allocations balancing valid inference with statistical efficiency. The oracle design is a covariate-dependent Neyman rule governed by unknown arm-conditional outcome variances. We investigate whether this…

View free PDFSource page