Steering Llama 2 with Contrastive Activation Addition
☆213May 23, 2024Updated last year
Alternatives and similar repositories for CAA
Users that are interested in CAA are comparing it to the libraries listed below
Sorting:
- LLM experiments done during SERI MATS - focusing on activation steering / interpreting activation spaces☆102Sep 21, 2023Updated 2 years ago
- Steering vectors for transformer language models in Pytorch / Huggingface☆140Feb 21, 2025Updated last year
- Improving Steering Vectors by Targeting Sparse Autoencoder Features☆27Nov 20, 2024Updated last year
- ☆209Oct 14, 2025Updated 4 months ago
- Algebraic value editing in pretrained language models☆69Nov 1, 2023Updated 2 years ago
- Code and results accompanying the paper "Refusal in Language Models Is Mediated by a Single Direction".☆355Jun 13, 2025Updated 8 months ago
- A resource repository for representation engineering in large language models☆148Nov 14, 2024Updated last year
- [ICLR 2025] General-purpose activation steering library☆145Sep 18, 2025Updated 5 months ago
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆247Feb 27, 2026Updated last week
- Materials for "Multi-property Steering of Large Language Models with Dynamic Activation Composition"☆14Nov 22, 2024Updated last year
- Official codebase for "Analyzing the Generalization and Reliability of Steering Vectors"☆19Dec 14, 2024Updated last year
- Open source replication of Anthropic's Crosscoders for Model Diffing☆64Oct 27, 2024Updated last year
- ☆30Aug 2, 2024Updated last year
- Training Sparse Autoencoders on Language Models☆1,233Feb 27, 2026Updated last week
- A tiny easily hackable implementation of a feature dashboard.☆15Oct 21, 2025Updated 4 months ago
- Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization☆42Jul 28, 2024Updated last year
- ☆36Apr 30, 2024Updated last year
- The nnsight package enables interpreting and manipulating the internals of deep learned models.☆836Updated this week
- ☆153Dec 30, 2025Updated 2 months ago
- Sparse Autoencoder for Mechanistic Interpretability☆292Jul 20, 2024Updated last year
- A collection of different ways to implement accessing and modifying internal model activations for LLMs☆20Oct 18, 2024Updated last year
- Sparse Autoencoder Training Library☆55May 1, 2025Updated 10 months ago
- Stanford NLP Python library for benchmarking the utility of LLM interpretability methods☆168Feb 22, 2026Updated last week
- A library for mechanistic interpretability of GPT-style language models☆3,133Updated this week
- ☆273Oct 1, 2024Updated last year
- ☆27Oct 6, 2024Updated last year
- This repository contains the code used for the experiments in the paper "Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity…☆30Oct 27, 2025Updated 4 months ago
- Sparsify transformers with SAEs and transcoders☆699Updated this week
- Code for In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steering☆198Feb 13, 2025Updated last year
- ☆100Aug 8, 2024Updated last year
- [NeurIPS25] Official repo for "Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning"☆42Oct 3, 2025Updated 5 months ago
- Function Vectors in Large Language Models (ICLR 2024)☆192Apr 17, 2025Updated 10 months ago
- Stanford NLP Python library for understanding and improving PyTorch models via interventions☆866Updated this week
- Representation Engineering: A Top-Down Approach to AI Transparency☆953Aug 14, 2024Updated last year
- Code for reproducing our paper "Not All Language Model Features Are Linear"☆84Nov 27, 2024Updated last year
- Code release for the paper "Style Vectors for Steering Generative Large Language Models", accepted to the Findings of the EACL 2024.☆36Sep 26, 2024Updated last year
- [NAACL'25 Oral] Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineering☆76Jan 16, 2026Updated last month
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆244Updated this week
- Inference-Time Intervention: Eliciting Truthful Answers from a Language Model☆572Jan 28, 2025Updated last year