Calibrating Lightweight Sparse Autoencoder Feature Steering
cs.CL, cs.AI
Submitted: 2025-06-14
Updated: 2026-09-18
Code: https://github.com/IBM/sae-steering
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Probing Classifiers: Promises, Shortcomings, and Advances
- Mechanistic Interpretability for AI Safety -- A Review
- Transformer Explainer: Learning LLM Transformers with Interactive Visual Explanation and Experimentation
- Sparse Autoencoders Find Highly Interpretable Features in Language Models
- Scaling and evaluating sparse autoencoders
- RAVEL: Evaluating Interpretability Methods on Disentangling Language Model Representations
- Scaling Laws for Neural Language Models
- Interpreting Attention Layer Outputs with Sparse Autoencoders
- Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2
- Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models
- The Hydra Effect: Emergent Self-repair in Language Model Computations
- Steering Llama 2 via Contrastive Activation Addition
- Towards Understanding Sycophancy in Language Models
- ZipIt! Merging Models from Different Tasks without Training
- LLM Circuit Analyses Are Consistent Across Training and Scale
- Steering Language Models With Activation Engineering
- Aligning Large Language Models with Human: A Survey
- Navigating the Landscape of Large Language Models: A Comprehensive Review and Analysis of Paradigms and Fine-Tuning Strategies
- Continual Learning for Large Language Models: A Survey
- Parameter-Efficient Fine-Tuning Methods for Pretrained Language Models: A Critical Review and Assessment
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