ArthurConmy / MishformerLensLinks
MishformerLens intends to be a drop-in replacement for TransformerLens that AST patches HuggingFace Transformers rather than implementing a custom, numerically inaccurate Transformer architecture.
☆10Updated last year
Alternatives and similar repositories for MishformerLens
Users that are interested in MishformerLens are comparing it to the libraries listed below
Sorting:
- A tiny easily hackable implementation of a feature dashboard.☆15Updated last month
- PyTorch and NNsight implementation of AtP* (Kramar et al 2024, DeepMind)☆20Updated 9 months ago
- Engine for collecting, uploading, and downloading model activations☆24Updated 6 months ago
- Sparse Autoencoder Training Library☆55Updated 5 months ago
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆221Updated 10 months ago
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆218Updated this week
- Steering vectors for transformer language models in Pytorch / Huggingface☆125Updated 7 months ago
- ☆36Updated last year
- Erasing concepts from neural representations with provable guarantees☆238Updated 8 months ago
- A library for training crosscoders☆11Updated 4 months ago
- ☆15Updated 3 months ago
- Code for reproducing our paper "Not All Language Model Features Are Linear"☆81Updated 10 months ago
- A library for efficient patching and automatic circuit discovery.☆77Updated 2 months ago
- ☆128Updated last year
- ☆73Updated last week
- Attribution-based Parameter Decomposition☆31Updated 4 months ago
- Mechanistic Interpretability Visualizations using React☆293Updated 10 months ago
- ☆109Updated 8 months ago
- ☆81Updated 7 months ago
- graphpatch is a library for activation patching on PyTorch neural network models.☆20Updated 8 months ago
- How do transformer LMs encode relations?☆55Updated last year
- ☆19Updated 6 months ago
- Code for my NeurIPS 2024 ATTRIB paper titled "Attribution Patching Outperforms Automated Circuit Discovery"☆42Updated last year
- Open source replication of Anthropic's Crosscoders for Model Diffing☆59Updated 11 months ago
- ☆244Updated last year
- ☆142Updated last month
- Modified to support crosscoder training.☆23Updated last week
- ☆53Updated 2 months ago
- Extract full next-token probabilities via language model APIs☆247Updated last year
- Unified access to Large Language Model modules using NNsight☆49Updated last week