IBM / sae-steeringLinks
Code to enable layer-level steering in LLMs using sparse auto encoders
☆25Updated 3 months ago
Alternatives and similar repositories for sae-steering
Users that are interested in sae-steering are comparing it to the libraries listed below
Sorting:
- Stanford NLP Python library for benchmarking the utility of LLM interpretability methods☆124Updated 2 months ago
- Code for reproducing our paper "Not All Language Model Features Are Linear"☆77Updated 9 months ago
- Code release for "Debating with More Persuasive LLMs Leads to More Truthful Answers"☆114Updated last year
- Code for the ICLR 2024 paper "How to catch an AI liar: Lie detection in black-box LLMs by asking unrelated questions"☆72Updated last year
- Sparse Autoencoder Training Library☆54Updated 4 months ago
- Algebraic value editing in pretrained language models☆65Updated last year
- ☆90Updated last year
- Improving Steering Vectors by Targeting Sparse Autoencoder Features☆24Updated 9 months ago
- Function Vectors in Large Language Models (ICLR 2024)☆177Updated 4 months ago
- ☆30Updated 6 months ago
- ☆72Updated last year
- This repository contains the code and data for the paper "SelfIE: Self-Interpretation of Large Language Model Embeddings" by Haozhe Chen,…☆51Updated 8 months ago
- Code repo for the model organisms and convergent directions of EM papers.☆26Updated last month
- LLM experiments done during SERI MATS - focusing on activation steering / interpreting activation spaces☆95Updated last year
- A library for efficient patching and automatic circuit discovery.☆76Updated last month
- Steering vectors for transformer language models in Pytorch / Huggingface☆122Updated 6 months ago
- ☆47Updated last year
- PaCE: Parsimonious Concept Engineering for Large Language Models (NeurIPS 2024)☆39Updated 9 months ago
- ☆187Updated 4 months ago
- [EMNLP 2025 Main] ConceptVectors Benchmark and Code for the paper "Intrinsic Evaluation of Unlearning Using Parametric Knowledge Traces"☆36Updated last week
- Open source replication of Anthropic's Crosscoders for Model Diffing☆58Updated 10 months ago
- [ICLR 2025] Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates (Oral)☆82Updated 10 months ago
- [ICLR'24] RAIN: Your Language Models Can Align Themselves without Finetuning☆97Updated last year
- [NeurIPS 2024 Spotlight] Code and data for the paper "Finding Transformer Circuits with Edge Pruning".☆59Updated 2 weeks ago
- Repo accompanying our paper "Do Llamas Work in English? On the Latent Language of Multilingual Transformers".☆78Updated last year
- [ICLR 2025] General-purpose activation steering library☆95Updated last month
- Replicating O1 inference-time scaling laws☆89Updated 9 months ago
- NeurIPS'24 - LLM Safety Landscape☆28Updated 6 months ago
- Improving Alignment and Robustness with Circuit Breakers☆228Updated 11 months ago
- We introduce EMMET and unify model editing with popular algorithms ROME and MEMIT.☆24Updated 8 months ago