timaeus-research / devinterpLinks
Tools for studying developmental interpretability in neural networks.
☆101Updated last month
Alternatives and similar repositories for devinterp
Users that are interested in devinterp are comparing it to the libraries listed below
Sorting:
- Mechanistic Interpretability Visualizations using React☆280Updated 8 months ago
- Resources for skilling up in AI alignment research engineering. Covers basics of deep learning, mechanistic interpretability, and RL.☆221Updated last week
- Notebooks accompanying Anthropic's "Toy Models of Superposition" paper☆129Updated 2 years ago
- ☆237Updated 10 months ago
- ☆126Updated last year
- 🧠 Starter templates for doing interpretability research☆73Updated 2 years ago
- Attribution-based Parameter Decomposition☆28Updated 2 months ago
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆211Updated 8 months ago
- (Model-written) LLM evals library☆18Updated 8 months ago
- Keeping language models honest by directly eliciting knowledge encoded in their activations.☆209Updated this week
- ☆32Updated last year
- Sparse Autoencoder for Mechanistic Interpretability☆258Updated last year
- ☆276Updated last year
- Sparse Autoencoder Training Library☆54Updated 3 months ago
- we got you bro☆36Updated last year
- ☆28Updated last year
- Machine Learning for Alignment Bootcamp☆77Updated 3 years ago
- Mechanistic Interpretability for Transformer Models☆51Updated 3 years ago
- Erasing concepts from neural representations with provable guarantees☆232Updated 6 months ago
- ☆63Updated 2 years ago
- A library for bridging Python and HTML/Javascript (via Svelte) for creating interactive visualizations☆14Updated last year
- METR Task Standard☆158Updated 6 months ago
- ☆122Updated last year
- 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
- Emergent world representations: Exploring a sequence model trained on a synthetic task☆188Updated 2 years ago
- The nnsight package enables interpreting and manipulating the internals of deep learned models.☆630Updated this week
- ☆327Updated this week
- ☆52Updated 9 months ago
- ☆16Updated last week
- ControlArena is a collection of settings, model organisms and protocols - for running control experiments.☆88Updated this week