mishajw / repengLinks
Experiments with representation engineering
☆12Updated last year
Alternatives and similar repositories for repeng
Users that are interested in repeng are comparing it to the libraries listed below
Sorting:
- A collection of different ways to implement accessing and modifying internal model activations for LLMs☆19Updated 9 months ago
- A library for efficient patching and automatic circuit discovery.☆74Updated 3 weeks ago
- LLM experiments done during SERI MATS - focusing on activation steering / interpreting activation spaces☆96Updated last year
- ☆125Updated last year
- ☆109Updated 3 weeks ago
- Steering vectors for transformer language models in Pytorch / Huggingface☆120Updated 5 months ago
- Open source replication of Anthropic's Crosscoders for Model Diffing☆57Updated 9 months ago
- 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
- ☆234Updated 10 months ago
- Algebraic value editing in pretrained language models☆65Updated last year
- Steering Llama 2 with Contrastive Activation Addition☆170Updated last year
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆202Updated this week
- Unified access to Large Language Model modules using NNsight☆38Updated 2 weeks ago
- ☆32Updated last year
- PyTorch and NNsight implementation of AtP* (Kramar et al 2024, DeepMind)☆18Updated 6 months ago
- ☆183Updated 3 weeks ago
- ☆51Updated 8 months ago
- Utilities for the HuggingFace transformers library☆70Updated 2 years ago
- NeuroSurgeon is a package that enables researchers to uncover and manipulate subnetworks within models in Huggingface Transformers☆41Updated 5 months ago
- Mechanistic Interpretability Visualizations using React☆273Updated 7 months ago
- Sparse Autoencoder Training Library☆54Updated 3 months ago
- Code for the ICLR 2024 paper "How to catch an AI liar: Lie detection in black-box LLMs by asking unrelated questions"☆71Updated last year
- Code for my NeurIPS 2024 ATTRIB paper titled "Attribution Patching Outperforms Automated Circuit Discovery"☆40Updated last year
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆210Updated 7 months ago
- Sparse probing paper full code.☆58Updated last year
- Code to reproduce key results accompanying "SAEs (usually) Transfer Between Base and Chat Models"☆12Updated last year
- Attribution-based Parameter Decomposition☆28Updated 2 months ago
- ☆122Updated last year
- Keeping language models honest by directly eliciting knowledge encoded in their activations.☆209Updated last week
- datasets from the paper "Towards Understanding Sycophancy in Language Models"☆86Updated last year