Thomas Hofmann

Affiliations

ETH Zurich

papers

711

total citations

papers (26)

Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers

NEURIPS 2023arXiv

citations

Convolutional Generation of Textured 3D Meshes

NEURIPS 2020arXiv

citations

Learning Generative Models of Textured 3D Meshes From Real-World Images

ICCV 2021arXiv

citations

Achieving a Better Stability-Plasticity Trade-Off via Auxiliary Networks in Continual Learning

CVPR 2023arXiv

citations

Precise characterization of the prior predictive distribution of deep ReLU networks

NEURIPS 2021arXiv

citations

Controlling Style and Semantics in Weakly-Supervised Image Generation

ECCV 2020arXiv

citations

Transformer Fusion with Optimal Transport

ICLR 2024arXiv

citations

Disentangling the Roles of Curation, Data-Augmentation and the Prior in the Cold Posterior Effect

NEURIPS 2021arXiv

citations

OpenFilter: A Framework to Democratize Research Access to Social Media AR Filters

NEURIPS 2022arXiv

citations

Recurrent Distance Filtering for Graph Representation Learning

ICML 2024arXiv

citations

Adversarial Training is a Form of Data-dependent Operator Norm Regularization

NEURIPS 2020arXiv

citations

LoRACLR: Contrastive Adaptation for Customization of Diffusion Models

CVPR 2025arXiv

citations

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

NEURIPS 2025arXiv

citations

On the Expressiveness and Length Generalization of Selective State Space Models on Regular Languages

AAAI 2025

citations

Batch normalization provably avoids ranks collapse for randomly initialised deep networks

NEURIPS 2020arXiv

citations

Mastering Spatial Graph Prediction of Road Networks

ICCV 2023arXiv

citations

The Importance of Being Lazy: Scaling Limits of Continual Learning

ICML 2025arXiv

citations

Navigating Scaling Laws: Compute Optimality in Adaptive Model Training

ICML 2024arXiv

citations

The Directionality of Optimization Trajectories in Neural Networks

ICLR 2025

citations

Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment

ICML 2025arXiv

citations

UIP2P: Unsupervised Instruction-based Image Editing via Edit Reversibility Constraint

ICCV 2025arXiv

citations

Thomas Hofmann

Affiliations

papers (26)

Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers

Convolutional Generation of Textured 3D Meshes

Learning Generative Models of Textured 3D Meshes From Real-World Images

Achieving a Better Stability-Plasticity Trade-Off via Auxiliary Networks in Continual Learning

Scaling MLPs: A Tale of Inductive Bias

Simplifying Transformer Blocks

Analytic Insights into Structure and Rank of Neural Network Hessian Maps

The Shaped Transformer: Attention Models in the Infinite Depth-and-Width Limit

A Language Model’s Guide Through Latent Space

Precise characterization of the prior predictive distribution of deep ReLU networks

Controlling Style and Semantics in Weakly-Supervised Image Generation

Transformer Fusion with Optimal Transport

Disentangling the Roles of Curation, Data-Augmentation and the Prior in the Cold Posterior Effect

OpenFilter: A Framework to Democratize Research Access to Social Media AR Filters

Recurrent Distance Filtering for Graph Representation Learning

Adversarial Training is a Form of Data-dependent Operator Norm Regularization

LoRACLR: Contrastive Adaptation for Customization of Diffusion Models

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

On the Expressiveness and Length Generalization of Selective State Space Models on Regular Languages

Batch normalization provably avoids ranks collapse for randomly initialised deep networks

Mastering Spatial Graph Prediction of Road Networks

The Importance of Being Lazy: Scaling Limits of Continual Learning

Navigating Scaling Laws: Compute Optimality in Adaptive Model Training

The Directionality of Optimization Trajectories in Neural Networks

Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment

UIP2P: Unsupervised Instruction-based Image Editing via Edit Reversibility Constraint

papers (26)

Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers

Convolutional Generation of Textured 3D Meshes

Learning Generative Models of Textured 3D Meshes From Real-World Images

Achieving a Better Stability-Plasticity Trade-Off via Auxiliary Networks in Continual Learning

Scaling MLPs: A Tale of Inductive Bias

Simplifying Transformer Blocks

Analytic Insights into Structure and Rank of Neural Network Hessian Maps

The Shaped Transformer: Attention Models in the Infinite Depth-and-Width Limit

A Language Model’s Guide Through Latent Space

Precise characterization of the prior predictive distribution of deep ReLU networks

Controlling Style and Semantics in Weakly-Supervised Image Generation

Transformer Fusion with Optimal Transport

Disentangling the Roles of Curation, Data-Augmentation and the Prior in the Cold Posterior Effect

OpenFilter: A Framework to Democratize Research Access to Social Media AR Filters

Recurrent Distance Filtering for Graph Representation Learning

Adversarial Training is a Form of Data-dependent Operator Norm Regularization

LoRACLR: Contrastive Adaptation for Customization of Diffusion Models

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

On the Expressiveness and Length Generalization of Selective State Space Models on Regular Languages

Batch normalization provably avoids ranks collapse for randomly initialised deep networks

Mastering Spatial Graph Prediction of Road Networks

The Importance of Being Lazy: Scaling Limits of Continual Learning

Navigating Scaling Laws: Compute Optimality in Adaptive Model Training

The Directionality of Optimization Trajectories in Neural Networks

Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment

UIP2P: Unsupervised Instruction-based Image Editing via Edit Reversibility Constraint