Image de couverture de livre

Entropy-Guided Attention for Private LLMs

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Jha, Nandan Kumar, Reagen, Brandon
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning Cryptography and Security
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

AERO: Entropy-Guided Framework for Private LLM Inference
par: Jha, Nandan Kumar, et autres
Publié: (2024)

DeepReShape: Redesigning Neural Networks for Efficient Private Inference
par: Jha, Nandan Kumar, et autres
Publié: (2023)

TruncFormer: Private LLM Inference Using Only Truncations
par: Yubeaton, Patrick, et autres
Publié: (2024)

Network and Compiler Optimizations for Efficient Linear Algebra Kernels in Private Transformer Inference
par: Garimella, Karthik, et autres
Publié: (2025)

A Random Matrix Theory Perspective on the Learning Dynamics of Multi-head Latent Attention
par: Jha, Nandan Kumar, et autres
Publié: (2025)

Differentially Private Attention Computation
par: Gao, Yeqi, et autres
Publié: (2023)

LLMStinger: Jailbreaking LLMs using RL fine-tuned LLMs
par: Jha, Piyush, et autres
Publié: (2024)

NerVE: Nonlinear Eigenspectrum Dynamics in LLM Feed-Forward Networks
par: Jha, Nandan Kumar, et autres
Publié: (2026)

Same Architecture, Different Capacity: Optimizer-Induced Spectral Scaling Laws
par: Jha, Nandan Kumar, et autres
Publié: (2026)

Spectral Scaling Laws in Language Models: How Effectively Do Feed-Forward Networks Use Their Latent Space?
par: Jha, Nandan Kumar, et autres
Publié: (2025)

ReLU's Revival: On the Entropic Overload in Normalization-Free Large Language Models
par: Jha, Nandan Kumar, et autres
Publié: (2024)

Osiris: A Systolic Approach to Accelerating Fully Homomorphic Encryption
par: Ebel, Austin, et autres
Publié: (2024)

HE-LRM: Efficient Private Embedding Lookups for Neural Inference Using Fully Homomorphic Encryption
par: Garimella, Karthik, et autres
Publié: (2025)

Is API Access to LLMs Useful for Generating Private Synthetic Tabular Data?
par: Swanberg, Marika, et autres
Publié: (2025)

Open LLMs are Necessary for Current Private Adaptations and Outperform their Closed Alternatives
par: Hanke, Vincent, et autres
Publié: (2024)

Private LoRA Fine-tuning of Open-Source LLMs with Homomorphic Encryption
par: Frery, Jordan, et autres
Publié: (2025)

An Attack to Break Permutation-Based Private Third-Party Inference Schemes for LLMs
par: Thomas, Rahul, et autres
Publié: (2025)

Adversarial Attacks on Locally Private Graph Neural Networks
par: Varun, Matta, et autres
Publié: (2026)

IF-GUIDE: Influence Function-Guided Detoxification of LLMs
par: Coalson, Zachary, et autres
Publié: (2025)

Pr$εε$mpt: Sanitizing Sensitive Prompts for LLMs
par: Chowdhury, Amrita Roy, et autres
Publié: (2025)

Privately Learning Decision Lists and a Differentially Private Winnow
par: Bun, Mark, et autres
Publié: (2026)

EinHops: Einsum Notation for Expressive Homomorphic Operations on RNS-CKKS Tensors
par: Garimella, Karthik, et autres
Publié: (2025)

Orion: A Fully Homomorphic Encryption Framework for Deep Learning
par: Ebel, Austin, et autres
Publié: (2023)

Private-RAG: Answering Multiple Queries with LLMs while Keeping Your Data Private
par: Wu, Ruihan, et autres
Publié: (2025)

Provable Differentially Private Computation of the Cross-Attention Mechanism
par: Ke, Yekun, et autres
Publié: (2024)

Scaling Laws for Differentially Private Language Models
par: McKenna, Ryan, et autres
Publié: (2025)

DictPFL: Efficient and Private Federated Learning on Encrypted Gradients
par: Xue, Jiaqi, et autres
Publié: (2025)

Private and Communication-Efficient Federated Learning based on Differentially Private Sketches
par: Zhang, Meifan, et autres
Publié: (2024)

DMM: Distributed Matrix Mechanism for Differentially-Private Federated Learning Based on Constant-Overhead Linear Secret Resharing
par: Bienstock, Alexander, et autres
Publié: (2024)

Learning with Locally Private Examples by Inverse Weierstrass Private Stochastic Gradient Descent
par: Dufraiche, Jean, et autres
Publié: (2026)

Breaking Bad: Interpretability-Based Safety Audits of State-of-the-Art LLMs
par: Agarwal, Krishiv, et autres
Publié: (2026)

Delving into Differentially Private Transformer
par: Ding, Youlong, et autres
Publié: (2024)

Privacy Profiles for Private Selection
par: Koskela, Antti, et autres
Publié: (2024)

Differentially Private Diffusion Models
par: Dockhorn, Tim, et autres
Publié: (2022)

AEGIS: Adversarial Entropy-Guided Immune System -- Thermodynamic State Space Models for Zero-Day Network Evasion Detection
par: Ferrel, Vickson
Publié: (2026)

MCP Safety Audit: LLMs with the Model Context Protocol Allow Major Security Exploits
par: Radosevich, Brandon, et autres
Publié: (2025)

On Design Principles for Private Adaptive Optimizers
par: Ganesh, Arun, et autres
Publié: (2025)

PAC-Private Responses with Adversarial Composition
par: Zhu, Xiaochen, et autres
Publié: (2026)

Locally Private Estimation with Public Features
par: Ma, Yuheng, et autres
Publié: (2024)

Differentially Private Conditional Independence Testing
par: Kalemaj, Iden, et autres
Publié: (2023)