evan-lloyd / graphpatch
graphpatch is a library for activation patching on PyTorch neural network models.
☆14Updated 2 months ago
Alternatives and similar repositories for graphpatch:
Users that are interested in graphpatch are comparing it to the libraries listed below
- MishformerLens intends to be a drop-in replacement for TransformerLens that AST patches HuggingFace Transformers rather than implementing…☆10Updated 6 months ago
- Mechanistic Interpretability for Transformer Models☆50Updated 2 years ago
- Sparse Autoencoder Training Library☆48Updated 5 months ago
- ☆121Updated last year
- Erasing concepts from neural representations with provable guarantees☆227Updated 3 months ago
- Replicating and dissecting the git-re-basin project in one-click-replication Colabs☆36Updated 2 years ago
- Universal Neurons in GPT2 Language Models☆27Updated 10 months ago
- Code for my NeurIPS 2024 ATTRIB paper titled "Attribution Patching Outperforms Automated Circuit Discovery"☆31Updated 10 months ago
- ☆33Updated 7 months ago
- Official repo for the paper "Weight-based Decomposition: A Case for Bilinear MLPs"☆20Updated 4 months ago
- Sparse and discrete interpretability tool for neural networks☆62Updated last year
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆197Updated 4 months ago
- ☆26Updated last year
- Understanding how features learned by neural networks evolve throughout training☆34Updated 6 months ago
- A library for efficient patching and automatic circuit discovery.☆63Updated this week
- ☆71Updated 2 months ago
- ☆51Updated 11 months ago
- ☆16Updated 2 months ago
- ☆90Updated 2 months ago
- ☆34Updated last year
- ☆219Updated 6 months ago
- ☆29Updated last week
- Simple implementation of muP, based on Spectral Condition for Feature Learning. The implementation is SGD only, dont use it for Adam☆76Updated 8 months ago
- How do transformer LMs encode relations?☆47Updated last year
- The simplest, fastest repository for training/finetuning medium-sized GPTs.☆105Updated 5 months ago
- ☆31Updated 3 months ago
- Notebooks accompanying Anthropic's "Toy Models of Superposition" paper☆120Updated 2 years ago
- ☆37Updated 2 months ago
- ☆95Updated last year
- Tiny re-implementation of MDM in style of LLaDA and nano-gpt speedrun☆48Updated last month