slavachalnev / SAE-TSLinks
Improving Steering Vectors by Targeting Sparse Autoencoder Features
☆24Updated 9 months ago
Alternatives and similar repositories for SAE-TS
Users that are interested in SAE-TS are comparing it to the libraries listed below
Sorting:
- ☆184Updated last month
- ☆114Updated last month
- Open source replication of Anthropic's Crosscoders for Model Diffing☆58Updated 9 months ago
- ☆53Updated 9 months ago
- A library for efficient patching and automatic circuit discovery.☆76Updated last month
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆206Updated last week
- Steering Llama 2 with Contrastive Activation Addition☆174Updated last year
- Evaluate interpretability methods on localizing and disentangling concepts in LLMs.☆53Updated 10 months ago
- Steering vectors for transformer language models in Pytorch / Huggingface☆122Updated 6 months ago
- [ICLR 2025] General-purpose activation steering library☆94Updated 3 weeks ago
- Sparse Autoencoder Training Library☆54Updated 3 months ago
- Algebraic value editing in pretrained language models☆65Updated last year
- Stanford NLP Python library for benchmarking the utility of LLM interpretability methods☆124Updated 2 months ago
- ☆162Updated 9 months ago
- Providing the answer to "How to do patching on all available SAEs on GPT-2?". It is an official repository of the implementation of the p…☆12Updated 7 months ago
- Code release for "Debating with More Persuasive LLMs Leads to More Truthful Answers"☆114Updated last year
- LLM experiments done during SERI MATS - focusing on activation steering / interpreting activation spaces☆95Updated last year
- ☆47Updated last year
- ☆238Updated 10 months ago
- Code to enable layer-level steering in LLMs using sparse auto encoders☆24Updated 3 months ago
- ☆103Updated 6 months ago
- A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity.☆76Updated 5 months ago
- ☆40Updated 7 months ago
- Using sparse coding to find distributed representations used by neural networks.☆262Updated last year
- Investigating the generalization behavior of LM probes trained to predict truth labels: (1) from one annotator to another, and (2) from e…☆28Updated last year
- ☆73Updated 3 months ago
- ☆122Updated last year
- ☆90Updated last year
- Code for my NeurIPS 2024 ATTRIB paper titled "Attribution Patching Outperforms Automated Circuit Discovery"☆40Updated last year
- ☆32Updated last year