Imagen de Portada

Beyond Input Activations: Identifying Influential Latents by Gradient Sparse Autoencoders

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Shu, Dong, Wu, Xuansheng, Zhao, Haiyan, Du, Mengnan, Liu, Ninghao
Formato:	Preprint
Publicado:	2025
Materias:	Machine Learning Artificial Intelligence Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models
por: Shu, Dong, et al.
Publicado: (2025)

Denoising Concept Vectors with Sparse Autoencoders for Improved Language Model Steering
por: Zhao, Haiyan, et al.
Publicado: (2025)

Enhancing LLM Steering through Sparse Autoencoder-Based Vector Refinement
por: Wang, Anyi, et al.
Publicado: (2025)

SAIF: A Sparse Autoencoder Framework for Interpreting and Steering Instruction Following of Language Models
por: He, Zirui, et al.
Publicado: (2025)

Comparative Analysis of Demonstration Selection Algorithms for LLM In-Context Learning
por: Shu, Dong, et al.
Publicado: (2024)

From Language Modeling to Instruction Following: Understanding the Behavior Shift in LLMs after Instruction Tuning
por: Wu, Xuansheng, et al.
Publicado: (2023)

Retrieval-enhanced Knowledge Editing in Language Models for Multi-Hop Question Answering
por: Shi, Yucheng, et al.
Publicado: (2024)

Beyond Single Concept Vector: Modeling Concept Subspace in LLMs with Gaussian Distribution
por: Zhao, Haiyan, et al.
Publicado: (2024)

Sparse Shift Autoencoders for Identifying Concepts from Large Language Model Activations
por: Joshi, Shruti, et al.
Publicado: (2025)

Exploring Multilingual Probing in Large Language Models: A Cross-Language Analysis
por: Li, Daoyang, et al.
Publicado: (2024)

Language Ranker: A Metric for Quantifying LLM Performance Across High and Low-Resource Languages
por: Li, Zihao, et al.
Publicado: (2024)

Jacobian Sparse Autoencoders: Sparsify Computations, Not Just Activations
por: Farnik, Lucy, et al.
Publicado: (2025)

Influential Language Data Selection via Gradient Trajectory Pursuit
por: Deng, Zhiwei, et al.
Publicado: (2024)

Rep2Text: Decoding Full Text from a Single LLM Token Representation
por: Zhao, Haiyan, et al.
Publicado: (2025)

Sparse Autoencoder Features for Classifications and Transferability
por: Gallifant, Jack, et al.
Publicado: (2025)

Fine-Grained Interpretation of Political Opinions in Large Language Models
por: Hu, Jingyu, et al.
Publicado: (2025)

Efficient Long-distance Latent Relation-aware Graph Neural Network for Multi-modal Emotion Recognition in Conversations
por: Shou, Yuntao, et al.
Publicado: (2024)

Incorporating Hierarchical Semantics in Sparse Autoencoder Architectures
por: Muchane, Mark, et al.
Publicado: (2025)

Self-Regularization with Sparse Autoencoders for Controllable LLM-based Classification
por: Wu, Xuansheng, et al.
Publicado: (2025)

Sparse Autoencoder Decomposition of Clinical Sequence Model Representations: Feature Complexity, Task Specialisation, and Mortality Prediction
por: Sainsbury, Chris, et al.
Publicado: (2026)

Beyond Adapter Retrieval: Latent Geometry-Preserving Composition via Sparse Task Projection
por: Jin, Pengfei, et al.
Publicado: (2024)

Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse Autoencoders
por: Chanin, David, et al.
Publicado: (2025)

Improving Steering Vectors by Targeting Sparse Autoencoder Features
por: Chalnev, Sviatoslav, et al.
Publicado: (2024)

Evaluating Adversarial Robustness of Concept Representations in Sparse Autoencoders
por: Li, Aaron J., et al.
Publicado: (2025)

LESS: Selecting Influential Data for Targeted Instruction Tuning
por: Xia, Mengzhou, et al.
Publicado: (2024)

Usable XAI: 10 Strategies Towards Exploiting Explainability in the LLM Era
por: Wu, Xuansheng, et al.
Publicado: (2024)

Beyond Gradient and Priors in Privacy Attacks: Leveraging Pooler Layer Inputs of Language Models in Federated Learning
por: Li, Jianwei, et al.
Publicado: (2023)

Diversity-driven Data Selection for Language Model Tuning through Sparse Autoencoder
por: Yang, Xianjun, et al.
Publicado: (2025)

AbsTopK: Rethinking Sparse Autoencoders For Bidirectional Features
por: Zhu, Xudong, et al.
Publicado: (2025)

Temporal Sparse Autoencoders: Leveraging the Sequential Nature of Language for Interpretability
por: Bhalla, Usha, et al.
Publicado: (2025)

Feature Hedging: Correlated Features Break Narrow Sparse Autoencoders
por: Chanin, David, et al.
Publicado: (2025)

SAEMark: Steering Personalized Multilingual LLM Watermarks with Sparse Autoencoders
por: Yu, Zhuohao, et al.
Publicado: (2025)

Rethinking Evaluation of Sparse Autoencoders through the Representation of Polysemous Words
por: Minegishi, Gouki, et al.
Publicado: (2025)

Steering LLMs? Actually, Sparse Autoencoders can outperform simple baselines
por: Jørgensen, Mikkel Godsk, et al.
Publicado: (2026)

Applying Large Language Models and Chain-of-Thought for Automatic Scoring
por: Lee, Gyeong-Geon, et al.
Publicado: (2023)

Montessori-Instruct: Generate Influential Training Data Tailored for Student Learning
por: Li, Xiaochuan, et al.
Publicado: (2024)

Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2
por: Lieberum, Tom, et al.
Publicado: (2024)

DLM-Scope: Mechanistic Interpretability of Diffusion Language Models via Sparse Autoencoders
por: Wang, Xu, et al.
Publicado: (2026)

Towards Understanding the Robustness of Sparse Autoencoders
por: Saiyed, Ahson, et al.
Publicado: (2026)

BRIDGE the Gap: Mitigating Bias Amplification in Automated Scoring of English Language Learners via Inter-group Data Augmentation
por: Wang, Yun, et al.
Publicado: (2026)