Thomas Hofmann

Heads collapse, features stay: Why Replay needs big buffers

Soft-Masked Diffusion Language Models

Barriers for Learning in an Evolving World: Mathematical Understanding of Loss of Plasticity

Cracking the Hessian: Closed-Form Hessian Spectra for Fundamental Neural Networks

On the Emergence of Induction Heads for In-Context Learning

Optimal Control Meets Flow Matching: A Principled Route to Multi-Subject Fidelity

Scaling Behavior of Discrete Diffusion Language Models

Model Fusion via Neuron Interpolation

Planner and Executor: Collaboration between Discrete Diffusion And Auto-regressive Models in Reasoning

The Non-Linear Representation Dilemma: Is Causal Abstraction Enough for Mechanistic Interpretability?

Structured Sparse Transition Matrices to Enable State Tracking in State-Space Models

The Directionality of Optimization Trajectories in Neural Networks

Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment

The Importance of Being Lazy: Scaling Limits of Continual Learning

UIP2P: Unsupervised Instruction-Based Image Editing via Cycle Edit Consistency

Generalized Interpolating Discrete Diffusion