TransluceAI / observatoryLinks
A toolkit for describing model features and intervening on those features to steer behavior.
☆197Updated 9 months ago
Alternatives and similar repositories for observatory
Users that are interested in observatory are comparing it to the libraries listed below
Sorting:
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆206Updated this week
- Open source interpretability artefacts for R1.☆157Updated 4 months ago
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆211Updated 8 months ago
- Steering vectors for transformer language models in Pytorch / Huggingface☆122Updated 6 months ago
- Mechanistic Interpretability Visualizations using React☆280Updated 8 months ago
- Code and results accompanying the paper "Refusal in Language Models Is Mediated by a Single Direction".☆257Updated 2 months ago
- Sparsify transformers with SAEs and transcoders☆608Updated this week
- ☆138Updated 4 months ago
- A simple unified framework for evaluating LLMs☆240Updated 4 months ago
- ☆111Updated last month
- Sparse Autoencoder for Mechanistic Interpretability☆258Updated last year
- Code for NeurIPS'24 paper 'Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalization'☆228Updated last month
- A Collection of Competitive Text-Based Games for Language Model Evaluation and Reinforcement Learning☆245Updated last week
- ☆162Updated 9 months ago
- ☆96Updated 3 months ago
- ☆237Updated 10 months ago
- Repository for the paper Stream of Search: Learning to Search in Language☆150Updated 6 months ago
- open source interpretability platform 🧠☆344Updated this week
- Open source replication of Anthropic's Crosscoders for Model Diffing☆58Updated 9 months ago
- ☆184Updated last month
- Extract full next-token probabilities via language model APIs☆247Updated last year
- Code for reproducing our paper "Not All Language Model Features Are Linear"☆77Updated 8 months ago
- ☆32Updated last year
- ☆126Updated last year
- ☆195Updated 5 months ago
- ☆52Updated 9 months ago
- A library for efficient patching and automatic circuit discovery.☆76Updated last month
- Notebooks accompanying Anthropic's "Toy Models of Superposition" paper☆129Updated 2 years ago
- Steering Llama 2 with Contrastive Activation Addition☆174Updated last year
- Stanford NLP Python library for benchmarking the utility of LLM interpretability methods☆124Updated last month