A Two-Stage Pretraining-Finetuning Framework for Treatment Effect Estimation with Unmeasured Confounding

Open in new window