Lorenzo Noci

33 papers A* 12A 1Journal 20
YearRankTypeTitle / Venue / Authors
2026 J jnl
Thinking into the Future: Latent Lookahead Training for Transformers.
CoRR
Lorenzo Noci, Gregor Bachmann, Seyed-Mohsen Moosavi-Dezfooli, Moin Nabi
2026 J jnl
Universal Dynamics of Warmup Stable Decay: understanding WSD beyond Transformers.
CoRR
Annalisa Belloni, Lorenzo Noci, Antonio Orvieto
2025 J jnl
Attention Retrieves, MLP Memorizes: Disentangling Trainable Components in the Transformer.
CoRR
Yihe Dong, Lorenzo Noci, Mikhail Khodak, Mufan Bill Li
2025 J jnl
Don't be lazy: CompleteP enables compute-efficient deep transformers.
CoRR
Nolan Dey, Bin Claire Zhang, Lorenzo Noci, Mufan Bill Li, Blake Bordelon, Shane Bergsma, Cengiz Pehlevan, Boris Hanin, Joel Hestness
2025 J jnl
On the Emergence of Induction Heads for In-Context Learning.
CoRR
Tiberiu Musat, Tiago Pimentel, Lorenzo Noci, Alessandro Stolfo, Mrinmaya Sachan, Thomas Hofmann
2025 A* conf
The Importance of Being Lazy: Scaling Limits of Continual Learning.
ICML
Jacopo Graldi, Alessandro Breccia, Giulia Lanzillotta, Thomas Hofmann, Lorenzo Noci
2025 J jnl
The Importance of Being Lazy: Scaling Limits of Continual Learning.
CoRR
Jacopo Graldi, Alessandro Breccia, Giulia Lanzillotta, Thomas Hofmann, Lorenzo Noci
2024 A* conf
Depthwise Hyperparameter Transfer in Residual Networks: Dynamics and Scaling Limit.
ICLR
Blake Bordelon, Lorenzo Noci, Mufan Bill Li, Boris Hanin, Cengiz Pehlevan
2024 A conf
How Good is a Single Basin?
AISTATS
Kai Lion, Lorenzo Noci, Thomas Hofmann, Gregor Bachmann
2024 J jnl
How Good is a Single Basin?
CoRR
Kai Lion, Lorenzo Noci, Thomas Hofmann, Gregor Bachmann
2024 A* conf
Super Consistency of Neural Network Landscapes and Learning Rate Transfer.
NeurIPS
Lorenzo Noci, Alexandru Meterez, Thomas Hofmann, Antonio Orvieto
2024 J jnl
Understanding and Minimising Outlier Features in Neural Network Training.
CoRR
Bobby He, Lorenzo Noci, Daniele Paliotta, Imanol Schlag, Thomas Hofmann
2024 A* conf
Understanding and Minimising Outlier Features in Transformer Training.
NeurIPS
Bobby He, Lorenzo Noci, Daniele Paliotta, Imanol Schlag, Thomas Hofmann
2024 J jnl
Why do Learning Rates Transfer? Reconciling Optimization and Scaling Limits for Deep Learning.
CoRR
Lorenzo Noci, Alexandru Meterez, Thomas Hofmann, Antonio Orvieto
2023 A* conf
Achieving a Better Stability-Plasticity Trade-off via Auxiliary Networks in Continual Learning.
CVPR
Sanghwan Kim, Lorenzo Noci, Antonio Orvieto, Thomas Hofmann
2023 J jnl
Achieving a Better Stability-Plasticity Trade-off via Auxiliary Networks in Continual Learning.
CoRR
Sanghwan Kim, Lorenzo Noci, Antonio Orvieto, Thomas Hofmann
2023 J jnl
Depthwise Hyperparameter Transfer in Residual Networks: Dynamics and Scaling Limit.
CoRR
Blake Bordelon, Lorenzo Noci, Mufan Bill Li, Boris Hanin, Cengiz Pehlevan
2023 J jnl
Disentangling Linear Mode-Connectivity.
CoRR
Gul Sena Altintas, Gregor Bachmann, Lorenzo Noci, Thomas Hofmann
2023 A* conf
Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers.
NeurIPS
Sotiris Anagnostidis, Dario Pavllo, Luca Biggio, Lorenzo Noci, Aurélien Lucchi, Thomas Hofmann
2023 J jnl
Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers.
CoRR
Sotiris Anagnostidis, Dario Pavllo, Luca Biggio, Lorenzo Noci, Aurélien Lucchi, Thomas Hofmann
2023 A* conf
The Curious Case of Benign Memorization.
ICLR
Sotiris Anagnostidis, Gregor Bachmann, Lorenzo Noci, Thomas Hofmann
2023 A* conf
The Shaped Transformer: Attention Models in the Infinite Depth-and-Width Limit.
NeurIPS
Lorenzo Noci, Chuning Li, Mufan Bill Li, Bobby He, Thomas Hofmann, Chris J. Maddison, Dan Roy
2023 J jnl
The Shaped Transformer: Attention Models in the Infinite Depth-and-Width Limit.
CoRR
Lorenzo Noci, Chuning Li, Mufan Bill Li, Bobby He, Thomas Hofmann, Chris J. Maddison, Daniel M. Roy
2022 A* conf
How Tempering Fixes Data Augmentation in Bayesian Neural Networks.
ICML
Gregor Bachmann, Lorenzo Noci, Thomas Hofmann
2022 J jnl
How Tempering Fixes Data Augmentation in Bayesian Neural Networks.
CoRR
Gregor Bachmann, Lorenzo Noci, Thomas Hofmann
2022 A* conf
Signal Propagation in Transformers: Theoretical Perspectives and the Role of Rank Collapse.
NeurIPS
Lorenzo Noci, Sotiris Anagnostidis, Luca Biggio, Antonio Orvieto, Sidak Pal Singh, Aurélien Lucchi
2022 J jnl
Signal Propagation in Transformers: Theoretical Perspectives and the Role of Rank Collapse.
CoRR
Lorenzo Noci, Sotiris Anagnostidis, Luca Biggio, Antonio Orvieto, Sidak Pal Singh, Aurélien Lucchi
2022 J jnl
The Curious Case of Benign Memorization.
CoRR
Sotiris Anagnostidis, Gregor Bachmann, Lorenzo Noci, Thomas Hofmann
2021 A* conf
Disentangling the Roles of Curation, Data-Augmentation and the Prior in the Cold Posterior Effect.
NeurIPS
Lorenzo Noci, Kevin Roth, Gregor Bachmann, Sebastian Nowozin, Thomas Hofmann
2021 J jnl
Disentangling the Roles of Curation, Data-Augmentation and the Prior in the Cold Posterior Effect.
CoRR
Lorenzo Noci, Kevin Roth, Gregor Bachmann, Sebastian Nowozin, Thomas Hofmann
2021 A* conf
Precise characterization of the prior predictive distribution of deep ReLU networks.
NeurIPS
Lorenzo Noci, Gregor Bachmann, Kevin Roth, Sebastian Nowozin, Thomas Hofmann
2021 J jnl
Precise characterization of the prior predictive distribution of deep ReLU networks.
CoRR
Lorenzo Noci, Gregor Bachmann, Kevin Roth, Sebastian Nowozin, Thomas Hofmann
2020 J jnl
Adversarial Learning for Debiasing Knowledge Graph Embeddings.
CoRR
Mario Arduini, Lorenzo Noci, Federico Pirovano, Ce Zhang, Yash Raj Shrestha, Bibek Paudel