Imagen de Portada

RAVEL: Evaluating Interpretability Methods on Disentangling Language Model Representations

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Huang, Jing, Wu, Zhengxuan, Potts, Christopher, Geva, Mor, Geiger, Atticus
Formato:	Preprint
Publicado:	2024
Materias:	Computation and Language Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Constructing Interpretable Features from Compositional Neuron Groups
por: Shafran, Or, et al.
Publicado: (2025)

ReFT: Representation Finetuning for Language Models
por: Wu, Zhengxuan, et al.
Publicado: (2024)

A Reply to Makelov et al. (2023)'s "Interpretability Illusion" Arguments
por: Wu, Zhengxuan, et al.
Publicado: (2024)

pyvene: A Library for Understanding and Improving PyTorch Models via Interventions
por: Wu, Zhengxuan, et al.
Publicado: (2024)

HyperSteer: Activation Steering at Scale with Hypernetworks
por: Sun, Jiuding, et al.
Publicado: (2025)

Mixing Mechanisms: How Language Models Retrieve Bound Entities In-Context
por: Gur-Arieh, Yoav, et al.
Publicado: (2025)

ScoNe: Benchmarking Negation Reasoning in Language Models With Fine-Tuning and In-Context Learning
por: She, Jingyuan Selena, et al.
Publicado: (2023)

Interpretability at Scale: Identifying Causal Mechanisms in Alpaca
por: Wu, Zhengxuan, et al.
Publicado: (2023)

AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders
por: Wu, Zhengxuan, et al.
Publicado: (2025)

HyperDAS: Towards Automating Mechanistic Interpretability with Hypernetworks
por: Sun, Jiuding, et al.
Publicado: (2025)

Enhancing Automated Interpretability with Output-Centric Feature Descriptions
por: Gur-Arieh, Yoav, et al.
Publicado: (2025)

From Directions to Regions: Decomposing Activations in Language Models via Local Geometry
por: Shafran, Or, et al.
Publicado: (2026)

Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Models
por: Ghandeharioun, Asma, et al.
Publicado: (2024)

How Do Transformers Learn Variable Binding in Symbolic Programs?
por: Wu, Yiwei, et al.
Publicado: (2025)

Evaluating Open-Source Sparse Autoencoders on Disentangling Factual Knowledge in GPT-2 Small
por: Chaudhary, Maheep, et al.
Publicado: (2024)

Demystifying Verbatim Memorization in Large Language Models
por: Huang, Jing, et al.
Publicado: (2024)

Backward Lens: Projecting Language Model Gradients into the Vocabulary Space
por: Katz, Shahar, et al.
Publicado: (2024)

Routers Learn the Geometry of Their Experts: Geometric Coupling in Sparse Mixture-of-Experts
por: Ahrac, Sagi, et al.
Publicado: (2026)

Recurrent Neural Networks Learn to Store and Generate Sequences using Non-Linear Representations
por: Csordás, Róbert, et al.
Publicado: (2024)

Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought
por: Boppana, Siddharth, et al.
Publicado: (2026)

How Causal Abstraction Underpins Computational Explanation
por: Geiger, Atticus, et al.
Publicado: (2025)

ReCOGS: How Incidental Details of a Logical Form Overshadow an Evaluation of Semantic Interpretation
por: Wu, Zhengxuan, et al.
Publicado: (2023)

Improved Representation Steering for Language Models
por: Wu, Zhengxuan, et al.
Publicado: (2025)

Language Models Encode Numbers Using Digit Representations in Base 10
por: Levy, Amit Arnold, et al.
Publicado: (2024)

Activation Steering via Generative Causal Mediation
por: Sankaranarayanan, Aruna, et al.
Publicado: (2026)

InnerThoughts: Disentangling Representations and Predictions in Large Language Models
por: Chételat, Didier, et al.
Publicado: (2025)

Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations
por: Geiger, Atticus, et al.
Publicado: (2023)

Internal Causal Mechanisms Robustly Predict Language Model Out-of-Distribution Behaviors
por: Huang, Jing, et al.
Publicado: (2025)

Disentangled Representation Learning with Large Language Models for Text-Attributed Graphs
por: Qin, Yijian, et al.
Publicado: (2023)

Don't Blame the Annotator: Bias Already Starts in the Annotation Instructions
por: Parmar, Mihir, et al.
Publicado: (2022)

Estimating Knowledge in Large Language Models Without Generating a Single Token
por: Gottesman, Daniela, et al.
Publicado: (2024)

Interpretable Discriminative Text Representations via Agreement and Label Disentanglement
por: Wang, Tong, et al.
Publicado: (2026)

Blackbox Model Provenance via Palimpsestic Membership Inference
por: Kuditipudi, Rohith, et al.
Publicado: (2025)

Disentangling MLP Neuron Weights in Vocabulary Space
por: Avrahamy, Asaf, et al.
Publicado: (2026)

Towards Interpreting Visual Information Processing in Vision-Language Models
por: Neo, Clement, et al.
Publicado: (2024)

PreFT: Prefill-only finetuning for efficient inference
por: Lanpouthakoun, Andrew, et al.
Publicado: (2026)

Mechanistically Interpretable Neural Encoding Reveals Fine-Grained Functional Selectivity in Human Visual Cortex
por: Grosbard, Idan Daniel, et al.
Publicado: (2026)

Evaluating Copyright Takedown Methods for Language Models
por: Wei, Boyi, et al.
Publicado: (2024)

RAVEL: Reasoning Agents for Validating and Evaluating LLM Text Synthesis
por: Feng, Andrew Zhuoer, et al.
Publicado: (2026)

When Can Transformers Count to n?
por: Yehudai, Gilad, et al.
Publicado: (2024)