AlignmentResearch / tuned-lens
Tools for understanding how transformer predictions are built layer-by-layer
☆480Updated 9 months ago
Alternatives and similar repositories for tuned-lens:
Users that are interested in tuned-lens are comparing it to the libraries listed below
- Mechanistic Interpretability Visualizations using React☆233Updated 3 months ago
- ☆211Updated 5 months ago
- Using sparse coding to find distributed representations used by neural networks.☆224Updated last year
- ☆263Updated last year
- ☆195Updated last year
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆189Updated 3 months ago
- This repository collects all relevant resources about interpretability in LLMs☆327Updated 4 months ago
- Sparsify transformers with SAEs and transcoders☆494Updated this week
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆161Updated this week
- ☆113Updated 7 months ago
- Stanford NLP Python library for understanding and improving PyTorch models via interventions☆718Updated 3 weeks ago
- Training Sparse Autoencoders on Language Models☆669Updated this week
- Mass-editing thousands of facts into a transformer memory (ICLR 2023)☆470Updated last year
- Keeping language models honest by directly eliciting knowledge encoded in their activations.☆197Updated this week
- ☆441Updated 8 months ago
- Erasing concepts from neural representations with provable guarantees☆226Updated last month
- ☆121Updated last year
- ☆255Updated last month
- Interpretability for sequence generation models 🐛 🔍☆409Updated 4 months ago
- Resources for skilling up in AI alignment research engineering. Covers basics of deep learning, mechanistic interpretability, and RL.☆206Updated last year
- Function Vectors in Large Language Models (ICLR 2024)☆149Updated this week
- The nnsight package enables interpreting and manipulating the internals of deep learned models.☆518Updated this week
- Sparse Autoencoder for Mechanistic Interpretability☆233Updated 8 months ago
- Editing Models with Task Arithmetic☆457Updated last year
- Inference-Time Intervention: Eliciting Truthful Answers from a Language Model☆514Updated last month
- Locating and editing factual associations in GPT (NeurIPS 2022)☆609Updated 11 months ago
- Steering Llama 2 with Contrastive Activation Addition☆129Updated 10 months ago
- Scaling Data-Constrained Language Models☆333Updated 6 months ago
- ☆61Updated 4 months ago
- Code to reproduce "Transformers Can Do Arithmetic with the Right Embeddings", McLeish et al (NeurIPS 2024)☆186Updated 9 months ago