Aryan Sharma

Publications

• Dissociating Decodability and Causal Use in Bracket-Sequence Transformers International Conference on Machine Learning (ICML), 2026 Compositional Learning Workshop Aryan Sharma, Cutter Dawes, Shivam Raval
• Hyperattentive Residuals ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2026 GALOP Workshop: Geometric Space, Architecture, and Learning Objective for Large Pre-Trained Models Grant Murphy, Neil He, Aryan Sharma, Menglin Yang, Rex Ying
• H-Probes: Extracting Hierarchical Structures From Latent Representations of Language Models Preprint (under review) Cutter Dawes*, Aryan Sharma*, Angelos Ioannis Lagos, Shivam Raval

* Equal contribution.