Gradients for Interventions and Activations for Detection: Targeted Feature Learning in Language Models
cs.CL, cs.LG
Submitted: 2026-09-26
Updated: 2026-09-26
Code: https://github.com/decoderesearch/SAELens
Terminology
Sources
- Improving Steering Vectors by Targeting Sparse Autoencoder Features
- The GRADIEND Python Package: An End-to-End System for Gradient-Based Feature Learning
- The Llama 3 Herd of Models
- A Unified Understanding and Evaluation of Steering Methods
- Steering LLMs? Actually, Sparse Autoencoders can outperform simple baselines
- Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)
- Locating and Editing Factual Associations in GPT
- Attention Is All You Need
- Representation Engineering: A Top-Down Approach to AI Transparency
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering