Accepted Posters

Posters will all be in A1 portrait format. Poster printing will not be provided, but can be arranged at local printing services such as:

Poster Session A: Foundations & Formal Theory

Benchmarking Neural Networks on Formal Language Classes

Alexandra Butoi, Nicola Lorenzon, Selim Jerad, Taiga Someya, Anej Svete, Ryan Cotterell, Brian DuSell
Non-Asymptotic Length Generalization

Thomas Chen, Tengyu Ma, Zhiyuan Li
Weighted Automata Learning with a Hard Inductive Bias

Breandan Considine
Neural Induction of Finite-State Transducers

Michael Ginn, Alexis Palmer, Mans Hulden
On the Learnability of Chain of Thought

Michael Hahn, Oliver Kraus, Yuekun Yao, Yash Sarrof, Alexander Koller
Learning Group Word Problems with Neural Networks

Hedayatullah Hakimi
Synthesis and Verification of Transformer Programs

Hongjian Jiang, Matthew Hague, Philipp Rümmer, Anthony Widjaja Lin
Logics with Neural Network Semantics

Caleb Schultz Kisby
C++-RASP: What Languages Can Transformers Length-Generalize at All?

Jiaoda Li, Ryan Cotterell
From Complexity to Clarity: A Three-Axis Study of Geometric Dynamics in LLM Hidden Representations

Xiang Li, Luzhe Sun, Matthew Walter, Jiawei Zhou
A Recall Tool Makes Linear RNNs Turing-Complete

Yanhong Li, William Merrill
Language Generation with Infinite Contamination

Anay Mehrotra, Grigoris Velegkas, Xifan Yu, Felix Zhou
The Infinity of Finite Numbers: Superfinite Capabilities in Finite Architectures

Naoki Negishi, Masaya Taniguchi, Keisuke Sakaguchi
Phase Transitions in the Learning of Regular Expressions: A Weight-Space Analysis

Prasanth
The Counting Power of Transformers

Marco Sälzer, Chris Köcher, Alexander Kozachinskiy, Georg Zetzsche, Anthony Widjaja Lin
Exact Expressivity of Unpadded Transformers

Anej Svete, Ashish Sabharwal, William Merrill
Revisiting Padded Transformer Expressivity: Which Architectural Choices Matter and Which Don't

Anej Svete, Ashish Sabharwal, William Merrill, Ryan Cotterell
Modeling Tool Use in Transformers Via Computation Oracles

Utkarsh Tiwari, Sai Soumya Nalli, Amit Deshpande, Michael Hahn
Evaluating Expressivity of Stack-Augmented Transformers across the Chomsky Hierarchy

Shunqi Wang
The Transformer Cookbook

Andy Yang, Christopher Watson, Anton Xue, Satwik Bhattamishra, Jose Llarena, William Merrill, Emile Dos Santos Ferreira, Anej Svete, David Chiang
Selective Irregularity in Japanese Verbs as a Formal Probe for Neural Sequence Models

Wen Zhang
(R)NNs too expressive? Neural Networks and the Subregular Hierarchy

Alan Zhou, Jane Li

Poster Session B: Mechanisms, Reasoning & Interpretability

On the Interaction Between Data Complexity and Scaling Laws

Vatsal Baherwani, Zixi Chen, Shikai Qiu, Andrew Gordon Wilson, Pavel Izmailov
Edge of Stochastic Stability: Revisiting the Edge of Stability for SGD

Pierfrancesco Beneventano, Arseniy Andreyev
Warm Starting State-Space Models with Automata Learning

William Peter Fishell, Sam Kouteili, Mark Paul Santolucito
Discovering Interpretable Algorithms by Decompiling Transformers to RASP

Xinting Huang, Aleksandra Bakalova, Satwik Bhattamishra, William Merrill, Michael Hahn
Transformers Provably Learn to Internalize Chain-of-Thought

Yixiao Huang, Hanlin Zhu, Zixuan Wang, Jiantao Jiao, Stuart Russell, Somayeh Sojoudi, Song Mei
CLAP: Carathéodory–Laplacian Attention Projection for Geometric Interpretation of Transformer Heads

Neev Justin
Autonomous Agents for Novel Architecture Discovery: Evaluating Expressivity, Efficiency, and Complexity

Brian Kitano
FOL-Traces: Verified First-Order Logic Reasoning Traces at Scale

Isabelle Lee, Sarah Liaw, Dani Yogatama
Can Interpretation Predict Model Behavior on Unseen Data?

Victoria R Li, Jenny Kaufmann, Martin Wattenberg, David Alvarez-Melis, Naomi Saphra
Hidden States as States: Discretizing Hidden-State Geometry in LLM Computation

Xiang Li, Luzhe Sun, Matthew Walter, Jiawei Zhou
Differentially Private Language Generation in the Limit

Anay Mehrotra, Grigoris Velegkas, Xifan Yu, Felix Zhou
Why Are Linear RNNs More Parallelizable?

William Merrill, Hongjian Jiang, Yanhong Li, Anthony Widjaja Lin, Ashish Sabharwal
LLM "Autoformalization" for Phonological Rule Induction

Abed Qaddoumi
On the Ability of Transformer to Verify Plans

Yash Sarrof, Yupei Du, Katharina Stein, Alexander Koller, Sylvie Thiebaux, Michael Hahn
Information Locality as an Inductive Bias for Neural Language Models

Taiga Someya, Anej Svete, Brian DuSell, Timothy J. O'Donnell, Mario Giulianelli, Ryan Cotterell
On the Reasoning Abilities of Masked Diffusion Language Models

Anej Svete, Ashish Sabharwal
Positional and Symbolic Attention in Transformers

Felipe Urrutia, Jorge Salas, Alexander Kozachinskiy, Cristian Buc Calderon, Hector Pasten, Cristobal Rojas
Emergence of Deferred Planning in Automata Embeddings for Task-Conditioned Reinforcement Learning

Marcell Vazquez-Chanlatte, Beyazit Yalcinkaya, Sanjit A. Seshia
w#w as an Inductive Bias in Pre-pretrained Language Models

Shunjie Wang
Provable Benefits of RLVR over SFT for Reasoning Models: Learning to Backtrack Efficiently

Stanley Wei, Juno Kim
Length Generalization Bounds for Transformers

Andy Yang, Pascal Bergsträßer, Georg Zetzsche, David Chiang, Anthony Widjaja Lin

Accepted Posters

Poster Session A: Foundations & Formal Theory

Benchmarking Neural Networks on Formal Language Classes

Non-Asymptotic Length Generalization

Weighted Automata Learning with a Hard Inductive Bias

Neural Induction of Finite-State Transducers

On the Learnability of Chain of Thought

Learning Group Word Problems with Neural Networks

Synthesis and Verification of Transformer Programs

Logics with Neural Network Semantics

C++-RASP: What Languages Can Transformers Length-Generalize at All?

From Complexity to Clarity: A Three-Axis Study of Geometric Dynamics in LLM Hidden Representations

A Recall Tool Makes Linear RNNs Turing-Complete

Language Generation with Infinite Contamination

The Infinity of Finite Numbers: Superfinite Capabilities in Finite Architectures

Phase Transitions in the Learning of Regular Expressions: A Weight-Space Analysis

The Counting Power of Transformers

Exact Expressivity of Unpadded Transformers

Revisiting Padded Transformer Expressivity: Which Architectural Choices Matter and Which Don't

Modeling Tool Use in Transformers Via Computation Oracles

Evaluating Expressivity of Stack-Augmented Transformers across the Chomsky Hierarchy

The Transformer Cookbook

Selective Irregularity in Japanese Verbs as a Formal Probe for Neural Sequence Models

(R)NNs too expressive? Neural Networks and the Subregular Hierarchy

Poster Session B: Mechanisms, Reasoning & Interpretability

On the Interaction Between Data Complexity and Scaling Laws

Edge of Stochastic Stability: Revisiting the Edge of Stability for SGD

Warm Starting State-Space Models with Automata Learning

Discovering Interpretable Algorithms by Decompiling Transformers to RASP

Transformers Provably Learn to Internalize Chain-of-Thought

CLAP: Carathéodory–Laplacian Attention Projection for Geometric Interpretation of Transformer Heads

Autonomous Agents for Novel Architecture Discovery: Evaluating Expressivity, Efficiency, and Complexity

FOL-Traces: Verified First-Order Logic Reasoning Traces at Scale

Can Interpretation Predict Model Behavior on Unseen Data?

Hidden States as States: Discretizing Hidden-State Geometry in LLM Computation

Differentially Private Language Generation in the Limit

Why Are Linear RNNs More Parallelizable?

LLM "Autoformalization" for Phonological Rule Induction

On the Ability of Transformer to Verify Plans

Information Locality as an Inductive Bias for Neural Language Models

On the Reasoning Abilities of Masked Diffusion Language Models

Positional and Symbolic Attention in Transformers

Emergence of Deferred Planning in Automata Embeddings for Task-Conditioned Reinforcement Learning

w#w as an Inductive Bias in Pre-pretrained Language Models

Provable Benefits of RLVR over SFT for Reasoning Models: Learning to Backtrack Efficiently

Length Generalization Bounds for Transformers