From Tokens to Policy: Causal and Interpretable Heterogeneous Treatment Effects Identification

arXiv:2606.17010 · cs.LG · Submitted 2026-06-15 · Read on arXiv

cs.LG

Submitted: 2026-06-15

Updated: 2026-09-30

Project page: https://www.riccardocadei.com/NEXIS

Terminology

• Heterogeneous Treatment Effect — 2× in this paper
• Interpretability — 1× in this paper · explained in 15 episodes

Sources

• Qwen2.5-VL Technical Report
• Causal Stability Selection
• Causal Rule Ensemble: Interpretable Discovery and Inference of Heterogeneous Treatment Effects
• Learning Multi-Level Features with Matryoshka Sparse Autoencoders
• Prediction-Powered Causal Inferences
• A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders
• Sparse Autoencoders Find Highly Interpretable Features in Language Models
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• Variable importance measures for heterogeneous treatment effects
• Foundation Models for Generalist Geospatial Artificial Intelligence
• Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Control
• Exploratory Causal Inference in SAEnce
• DINOv2: Learning Robust Visual Features without Supervision
• Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models
• Measuring Variable Importance in Heterogeneous Treatment Effects with Confidence
• The Linear Representation Hypothesis and the Geometry of Large Language Models
• SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features
• The Third Pillar of Causal Analysis? A Measurement Perspective on Causal Representations

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks