Thomas Hofmann
Full Professor
Leads the Data Analytics Lab, studying deep learning theory, the optimization and generalization of neural networks, and natural language understanding with large language models.
Thomas Hofmann's lab at ETH Zurich works on learning theory & optimization, large language models and data management & big data. 79 papers since 2023, 58 at top venues, mostly ICML and NeurIPS. 7 PhD students. 4 alumni.
At a glance
- papers, 2023–2026
- 21/ year
- at T1 venues
- 73%
- PhD students
- 7
- MS students
- 0
- alumni listed
- 4
Papers by yearfrom the lab homepage, 2023–2026
- 0 papers2026 · T1 0 · collecting
- 26 papers2025 · T1 23
- 31 papers2024 · T1 19
- 22 papers2023 · T1 16
T1 venuesOtherVenues: ICML, NeurIPS, ICLR, EMNLP, ACL
Research focuspaper titles that use each topic's words
- Large language models19 papers
- Learning theory & optimization13 papers
- generalization2 papers
- Data management & big data0 papers
- data analytics0 papers
Paper list
From the lab homepage.
202526 papers
- CVPR (T1)Autoregressive Distillation of Diffusion Transformers
- ICML (T1)Avoiding spurious sharpness minimization broadens applicability of SAM
- ACL (T1)Causal Estimation of Tokenisation Bias
- ICCV (T1)CLoRA: Contrastive Test-Time Composition of Multiple LoRA Models for Image Generation
- EMNLP (T1)Convergence and Divergence of Language Models under Different Random Seeds
- NeurIPS (T1)Don't be lazy: CompleteP enables compute-efficient deep transformers
- AISTATS (T1)Emergence of Globally Attracting Fixed Points in Deep Neural Networks With Nonlinear Activations
- CVPR (T1)FlexiDiT: Your Diffusion Transformer Can Easily Generate High-Quality Samples with Less Compute
- ICML (T1)Generalized Interpolating Discrete Diffusion
- NeurIPS (T1)Generalized Linear Mode Connectivity for Transformers
- ICML PUT WorkshopJEDI: The Force of Jensen-Shannon Divergence in Disentangling Diffusion Models
- ICLR (T1)Judge Decoding: Faster Speculative Sampling Requires Going Beyond Model Alignment
202431 papers
- ICML (T1)A Language Model's Guide Through Latent Space
- CognitionAn information-theoretic analysis of targeted regressions during reading
- ACL (T1)Causal Estimation of Memorisation Profiles
- ICML HiLD workshopClosed form of the Hessian spectrum for some Neural Networks
- CVPR (T1)CONFORM: Contrast is All You Need For High-Fidelity Text-to-Image Diffusion Models
- NeurIPS Continual FoMo WorkshopExploring the Limits of Feature Learning in Continual Learning
- ICML TF2M workshopHallmarks of Optimization Trajectories in Neural Networks and LLMs: Directional Redundancy and Exploration
- AISTATS (T1)How Good is a Single Basin?
- COLMHow Susceptible are LLMs to Influence in Prompts?
- EMNLP (T1)How to Compute the Probability of a Word
- ICML HiLD Workshop on High-dimensional Learning DynamicsInterpolated-MLPs: Controllable Inductive Bias
- TACL (T1)Investigating Critical Period Effects in Language Acquisition through Neural Language Models
202322 papers
- CVPR (T1)Achieving a Better Stability-Plasticity Trade-off via Auxiliary Networks in Continual Learning
- EMNLP (T1)An Exploration of Left-Corner Transformations
- ICML (T1)An SDE for Modeling SAM: Theory and Insights
- NeurIPS (T1)Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers
- NeurIPS M3L workshopEscaping Random Teacher Initialization Enhances Signal Propagation and Representations
- AISTATS (T1)Explicit Regularization in Overparametrized Models via Noise Injection
- NeurIPS SyntheticData4MLHarnessing Synthetic Datasets: The Role of Shape Bias in Deep Neural Network Generalization
- EMNLP (T1)Language Model Quality Correlates with Psychometric Predictive Power in Multiple Languages
- ICCV AI3DCCLatentSwap3D: Semantic Edits on 3D Image GAN
- ICCV (T1)Mastering Spatial Graph Prediction of Road Networks
- Journal of Physics A: Mathematical and TheoreticalMean First Exit Times of Ornstein-Uhlenbeck Processes in High Dimensions
- BMVCMulti-CLIP: Contrastive Vision-Language Pre-training for Question Answering tasks in 3D Scenes
Where alumni went4 listed
Apply to ETH ZurichRolling
Not announced yet
Computer Science Doctorate (D-INFK)
- Round
- Rolling (no deadline)
- Opens
- —
- GRE
- See program page
- English
- Proficiency in written and spoken English
- Fee
- See program page
Confirmed on the program page.
Similar labsnearest by research summary
- USUC Berkeley
Kannan Ramchandran
- Signal processing & information theory
- USCMU
Yonatan Bisk
- Multimodal & grounded language
- CNTsinghua
Yuantao Gu
- Learning theory & optimization
- USMichigan
Samet Oymak
- Learning theory & optimization
- USUCLA
Cho-Jui Hsieh
- Trustworthy & safe ML
- GBOxford
Giuseppe De Giacomo
- Reasoning & neuro-symbolic AI
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.