The field of treatment effect estimation and document retrieval is witnessing significant advancements, driven by innovations in conformal inference, multi-layer representations, and novel training objectives. Researchers are exploring new methods to improve the estimation of treatment effects, such as overlap-adaptive regularization, which addresses the issue of low overlap in conditional average treatment effect estimation. Meanwhile, document retrieval is being enhanced through the development of more efficient and effective reranking models, including those that utilize multi-layer representations, synthetic data, and LLM-based supervision. Noteworthy papers in this area include: A Systematic Review of Conformal Inference Procedures for Treatment Effect Estimation, which provides a comprehensive overview of conformal prediction methods for treatment effect estimation. Investigating Multi-layer Representations for Dense Passage Retrieval, which proposes a novel approach to representing documents using multi-layer representations. Overlap-Adaptive Regularization for Conditional Average Treatment Effect Estimation, which introduces a new regularization technique to improve the performance of meta-learners in low-overlap regions. jina-reranker-v3, which presents a compact architecture that achieves state-of-the-art performance in document reranking. Optimizing What Matters: AUC-Driven Learning for Robust Neural Retrieval, which introduces a new training objective that maximizes the Mann-Whitney U statistic, leading to better-calibrated and more discriminative retrievers.