Copertina

Curvature Tuning: Provable Training-free Model Steering From a Single Parameter

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Hu, Leyang, Gamba, Matteo, Balestriero, Randall
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Eidetic Learning: an Efficient and Provable Solution to Catastrophic Forgetting
di: Dronen, Nicholas, et al.
Pubblicazione: (2025)

LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics
di: Balestriero, Randall, et al.
Pubblicazione: (2025)

Task Priors: Enhancing Model Evaluation by Considering the Entire Space of Downstream Tasks
di: Patel, Niket, et al.
Pubblicazione: (2025)

No Location Left Behind: Measuring and Improving the Fairness of Implicit Representations for Earth Data
di: Cai, Daniel, et al.
Pubblicazione: (2025)

SAFE: A Novel Approach to AI Weather Evaluation through Stratified Assessments of Forecasts over Earth
di: Masi, Nick, et al.
Pubblicazione: (2025)

ALLoRA: Adaptive Learning Rate Mitigates LoRA Fatal Flaws
di: Huang, Hai, et al.
Pubblicazione: (2024)

Post-Hoc Guidance for Consistency Models by Joint Flow Distribution Learning
di: Hsu, Chia-Hong, et al.
Pubblicazione: (2026)

Fast and Exact Enumeration of Deep Networks Partitions Regions
di: Balestriero, Randall, et al.
Pubblicazione: (2024)

The Fair Language Model Paradox
di: Pinto, Andrea, et al.
Pubblicazione: (2024)

Joint Embedding vs Reconstruction: Provable Benefits of Latent Space Prediction for Self Supervised Learning
di: Van Assel, Hugues, et al.
Pubblicazione: (2025)

Characterizing Large Language Model Geometry Helps Solve Toxicity Detection and Generation
di: Balestriero, Randall, et al.
Pubblicazione: (2023)

Semantic Tube Prediction: Beating LLM Data Efficiency with JEPA
di: Huang, Hai, et al.
Pubblicazione: (2026)

Variance Covariance Regularization Enforces Pairwise Independence in Self-Supervised Representations
di: Mialon, Grégoire, et al.
Pubblicazione: (2022)

Occam's Razor for Self Supervised Learning: What is Sufficient to Learn Good Representations?
di: Ibrahim, Mark, et al.
Pubblicazione: (2024)

Your Attention Matters: to Improve Model Robustness to Noise and Spurious Correlations
di: Tamayo-Rousseau, Camilo, et al.
Pubblicazione: (2025)

Learning by Reconstruction Produces Uninformative Features For Perception
di: Balestriero, Randall, et al.
Pubblicazione: (2024)

Continual Fine-Tuning with Provably Accurate and Parameter-Free Task Retrieval
di: Le, Hang Thi-Thuy, et al.
Pubblicazione: (2026)

The Geometric Structure of Models Learning Sparse Data
di: Walker, Thomas, et al.
Pubblicazione: (2026)

Self-Supervised Anomaly Detection in the Wild: Favor Joint Embeddings Methods
di: Otero, Daniel, et al.
Pubblicazione: (2024)

GrokAlign: Geometric Characterisation and Acceleration of Grokking
di: Walker, Thomas, et al.
Pubblicazione: (2025)

The Linear Centroids Hypothesis: Features as Directions Learned by Local Experts
di: Walker, Thomas, et al.
Pubblicazione: (2026)

Position: An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning Research
di: Reizinger, Patrik, et al.
Pubblicazione: (2025)

stable-pretraining-v1: Foundation Model Research Made Simple
di: Balestriero, Randall, et al.
Pubblicazione: (2025)

FastDINOv2: Frequency Based Curriculum Learning Improves Robustness and Training Speed
di: Zhang, Jiaqi, et al.
Pubblicazione: (2025)

SplInterp: Improving our Understanding and Training of Sparse Autoencoders
di: Budd, Jeremy, et al.
Pubblicazione: (2025)

Deep Networks Always Grok and Here is Why
di: Humayun, Ahmed Imtiaz, et al.
Pubblicazione: (2024)

On the Geometry of Deep Learning
di: Balestriero, Randall, et al.
Pubblicazione: (2024)

PrAg-PO: Prompt Augmented Policy Optimization for Robust and Diverse Mathematical Reasoning
di: Lu, Wenquan, et al.
Pubblicazione: (2026)

Beyond [cls]: Exploring the true potential of Masked Image Modeling representations
di: Przewięźlikowski, Marcin, et al.
Pubblicazione: (2024)

Linear Independence of Generalized Neurons and Related Functions
di: Zhang, Leyang
Pubblicazione: (2024)

Mitigating over-exploration in latent space optimization using LES
di: Ronen, Omer, et al.
Pubblicazione: (2024)

LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels
di: Maes, Lucas, et al.
Pubblicazione: (2026)

Towards Decentralized and Sustainable Foundation Model Training with the Edge
di: Xue, Leyang, et al.
Pubblicazione: (2025)

Curvature-Guided LoRA: Steering in the pretrained NTK subspace
di: Zheng, Frédéric, et al.
Pubblicazione: (2026)

Provably Protecting Fine-Tuned LLMs from Training Data Extraction while Preserving Utility
di: Segal, Tom, et al.
Pubblicazione: (2026)

SplineCam: Exact Visualization and Characterization of Deep Network Geometry and Decision Boundaries
di: Humayun, Ahmed Imtiaz, et al.
Pubblicazione: (2023)

LoRA Users Beware: A Few Spurious Tokens Can Manipulate Your Finetuned Model
di: Salles, Marcel Mateos, et al.
Pubblicazione: (2025)

EmbedOR: Provable Cluster-Preserving Visualizations with Curvature-Based Stochastic Neighbor Embeddings
di: Saidi, Tristan Luca, et al.
Pubblicazione: (2025)

Task-Aware Parameter-Efficient Fine-Tuning of Large Pre-Trained Models at the Edge
di: Hu, Senkang, et al.
Pubblicazione: (2025)

Provably Robust Training of Quantum Circuit Classifiers Against Parameter Noise
di: Tecot, Lucas, et al.
Pubblicazione: (2025)