1. Barriers to Universal Reasoning With Transformers (And How to Overcome Them)

   Oliver Kraus*, Yash Sarrof*, Yuekun Yao, Alexander Koller, and 1 more author

   COLM 2026

   PDF Code

2. On the Ability of Transformers to Verify Plans

   Yash Sarrof, Yupei Du, Katharina Stein, Alexander Koller, and 2 more authors

   ICML 2026

   PDF Code

3. Born a Transformer–Always a Transformer? On the effect of Pretraining on Architectural Abilities

   Mayank Jobanputra, Yana Veitsman, Yash Sarrof, Aleksandra Bakalova, and 3 more authors

   NeurIPS 2025

   PDF Code

4. A Formal Framework for Understanding Length Generalization in Transformers

   Xinting Huang, Andy Yang, Satwik Bhattamishra, Yash Sarrof, and 4 more authors

   ICLR 2025

   PDF Code

5. The Expressive Capacity of State Space Models: A Formal Language Perspective

   Yash Sarrof, Yana Veitsman, and Michael Hahn

   NeurIPS 2024

   PDF Code

6. Homophonic Pun Generation in Code Mixed Hindi English

   Yash Sarrof

   CHum Computational Humour Workshop, COLING 2025

   PDF Code