Scholar
Anej Svete
Google Scholar ID: 9ezEOeUAAAAJ
ETH Zurich
Citations & Impact
Citations
181
H-index
8
i10-index
7
Publications
20
Co-authors
37
list available
Publications
Cited
0
Resume (English only)
Academic Achievements
- Published multiple papers at top-tier venues including ACL, ICLR, EMNLP, NAACL, and NeurIPS, such as:
- “The Exact Expressive Power of Fixed-Precision Looped Padded Transformers” (arXiv)
- “Information Locality as an Inductive Bias for Neural Language Models” (ACL 2025)
- “Unique Hard Attention: A Tale of Two Sides” (ACL 2025)
- “Gumbel Counterfactual Generation From Language Models” (ICLR 2025)
- “Training Neural Networks as Recognizers of Formal Languages” (ICLR 2025)
- “A Probability-Quality Trade-off in Aligned Language Models and its Relation to Sampling Adaptors” (EMNLP 2024)
- “Can Transformers Learn n-gram Language Models?” (ACL 2024)
- “On Efficiently Representing Regular Languages as RNNs” (ACL 2024 Findings)
- “An L* Algorithm for Deterministic Weighted Regular Languages” (ACL 2024)
- “On the Representational Capacity of Neural Language Models with Chain-of-Thought Reasoning” (ACL 2024)
- “On Affine Homotopy between Language Encoders” (NeurIPS 2024)
- “What Languages are Easy to Language-Model? A Perspective from Learning Probabilistic Regular Languages” (ACL 2024)
- “Transformers Can Represent n-gram Language Models” (NAACL 2024)
- “Lower Bounds on the Expressivity of Recurrent Neural Language Models” (NAACL 2024)
- “The Role of n-gram Smoothing in the Age of Neural Networks” (NAACL 2024)
- Invited talk at NeurIPS 2025 Workshop on Principles of Generative Modeling (December 2025)
- Organizing tutorial on The Underlying Logic of Language Models at ICML 2025 (July 2025)
- Organizing tutorial on Computational Expressivity of Neural Language Models at ACL 2024 (August 2024)
Background
- PhD Student in Natural Language Processing at ETH Zürich
- Research at the intersection of formal language theory and modern language models
- Investigates the capabilities and limitations of neural networks (e.g., Transformers): what problems they can solve, which aspects of language they capture, and whether they can truly 'reason'
- Student Researcher at the Allen Institute for AI (Ai2) since Summer 2025
- Collaborating with Ashish Sabharwal on reasoning and problem-solving in language models
- Co-advised by Prof. Ryan Cotterell and Prof. Valentina Boeva
- Co-organizes the Formal Languages and Neural Networks (FLaNN) Seminar
