Image de couverture de livre

Resting Neurons, Active Insights: Robustifying Activation Sparsity in LLMs via Spontaneity

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Xu, Haotian, Yang, Jiannan, Gao, Tian, Weng, Tsui-Wei, Ma, Tengfei
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Graph Concept Bottleneck Models
par: Xu, Haotian, et autres
Publié: (2025)

Linear Explanations for Individual Neurons
par: Oikarinen, Tuomas, et autres
Publié: (2024)

Evaluating Neuron Explanations: A Unified Framework with Sanity Checks
par: Oikarinen, Tuomas, et autres
Publié: (2025)

Faithful and Stable Neuron Explanations for Trustworthy Mechanistic Interpretability
par: Yan, Ge, et autres
Publié: (2025)

When Structure Doesn't Help: LLMs Do Not Read Text-Attributed Graphs as Effectively as We Expected
par: Xu, Haotian, et autres
Publié: (2025)

Joint Training Across Multiple Activation Sparsity Regimes
par: Wang, Haotian
Publié: (2026)

Self-Supervised Learning on Molecular Graphs: A Systematic Investigation of Masking Design
par: Yang, Jiannan, et autres
Publié: (2025)

Abstracted Shapes as Tokens -- A Generalizable and Interpretable Model for Time-series Classification
par: Wen, Yunshi, et autres
Publié: (2024)

Interpreting Neurons in Deep Vision Networks with Language Models
par: Bai, Nicholas, et autres
Publié: (2024)

Efficient Imputation for Patch-based Missing Single-cell Data via Cluster-regularized Optimal Transport
par: Liu, Yuyu, et autres
Publié: (2026)

Beyond Top Activations: Efficient and Reliable Crowdsourced Evaluation of Automated Interpretability
par: Oikarinen, Tuomas, et autres
Publié: (2025)

To 2:4 Sparsity and Beyond: Neuron-level Activation Function to Accelerate LLM Pre-Training
par: Madhyastha, Meghana, et autres
Publié: (2026)

Breaking the Barrier: Enhanced Utility and Robustness in Smoothed DRL Agents
par: Sun, Chung-En, et autres
Publié: (2024)

Neuron Activation Coverage: Rethinking Out-of-distribution Detection and Generalization
par: Liu, Yibing, et autres
Publié: (2023)

Towards the Connection between Activation Sparsity and Flat Minima
par: Peng, Ze, et autres
Publié: (2026)

Distillation Robustifies Unlearning
par: Lee, Bruce W., et autres
Publié: (2025)

Wasserstein Distances, Neuronal Entanglement, and Sparsity
par: Sawmya, Shashata, et autres
Publié: (2024)

ProTransformer: Robustify Transformers via Plug-and-Play Paradigm
par: Hou, Zhichao, et autres
Publié: (2024)

Iterative Self-Tuning LLMs for Enhanced Jailbreaking Capabilities
par: Sun, Chung-En, et autres
Publié: (2024)

RAT: Boosting Misclassification Detection Ability without Extra Data
par: Yan, Ge, et autres
Publié: (2025)

Interpretability-Guided Test-Time Adversarial Defense
par: Kulkarni, Akshay, et autres
Publié: (2024)

DuoGPT: Training-free Dual Sparsity through Activation-aware Pruning in LLMs
par: Yin, Ruokai, et autres
Publié: (2025)

Understanding Fixed Predictions via Confined Regions
par: Lawless, Connor, et autres
Publié: (2025)

R-Sparse: Rank-Aware Activation Sparsity for Efficient LLM Inference
par: Zhang, Zhenyu, et autres
Publié: (2025)

Týr-the-Pruner: Structural Pruning LLMs via Global Sparsity Distribution Optimization
par: Li, Guanchen, et autres
Publié: (2025)

Explore Activation Sparsity in Recurrent LLMs for Energy-Efficient Neuromorphic Computing
par: Knunyants, Ivan, et autres
Publié: (2025)

ThinkEdit: Interpretable Weight Editing to Mitigate Overly Short Thinking in Reasoning Models
par: Sun, Chung-En, et autres
Publié: (2025)

Effective Skill Unlearning through Intervention and Abstention
par: Li, Yongce, et autres
Publié: (2025)

Crafting Large Language Models for Enhanced Interpretability
par: Sun, Chung-En, et autres
Publié: (2024)

Reinforcement Learning With Sparse-Executing Actions via Sparsity Regularization
par: Pang, Jing-Cheng, et autres
Publié: (2021)

Robustifying Conditional Portfolio Decisions via Optimal Transport
par: Nguyen, Viet Anh, et autres
Publié: (2021)

VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance
par: Srivastava, Divyansh, et autres
Publié: (2024)

Task-Specific Data Selection for Instruction Tuning via Monosemantic Neuronal Activations
par: Ma, Da, et autres
Publié: (2025)

Robustifying and Boosting Training-Free Neural Architecture Search
par: He, Zhenfeng, et autres
Publié: (2024)

Deep Neural Network Initialization with Sparsity Inducing Activations
par: Price, Ilan, et autres
Publié: (2024)

Prediction without Preclusion: Recourse Verification with Reachable Sets
par: Kothari, Avni, et autres
Publié: (2023)

Provably Robust Conformal Prediction with Improved Efficiency
par: Yan, Ge, et autres
Publié: (2024)

Robustifying Point Cloud Networks by Refocusing
par: Levi, Meir Yossef, et autres
Publié: (2023)

Zeroth-Order Fine-Tuning of LLMs with Extreme Sparsity
par: Guo, Wentao, et autres
Publié: (2024)

SAU: Sparsity-Aware Unlearning for LLMs via Gradient Masking and Importance Redistribution
par: Wang, Yuze, et autres
Publié: (2026)