Cover Image

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

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Hu, Leyang, Gamba, Matteo, Balestriero, Randall
Format:	Preprint
Published:	2025
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

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

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

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

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

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

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

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

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

The Fair Language Model Paradox
by: Pinto, Andrea, et al.
Published: (2024)

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

Linear Independence of Generalized Neurons and Related Functions
by: Zhang, Leyang
Published: (2024)

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

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

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

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

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

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

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

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

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

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