AI Safety & Interpretability Lab

Popular repositories Loading

1. arbiter arbiter Public

   Run HuggingFace models through freeform questions and judge responses with an LLM.

   Python 5 1

2. psychological-safety psychological-safety Public

   Python 2

3. rapidrouge rapidrouge Public

   Drop-in rouge-score replacement with a bit-parallel ROUGE-L

   Python 2

4. diffing-toolkit diffing-toolkit Public

   Forked from science-of-finetuning/diffing-toolkit

   A toolkit that provides a range of model diffing techniques including a UI to visualize them interactively.

   Python 1 1

5. aisilab.github.io aisilab.github.io Public

   Website of the AI Safety & Interpretability Lab at SDU

   HTML 1 1

6. model_organism_trainer model_organism_trainer Public

   One-file Unsloth LoRA finetuner for taboo model organisms

   Python 1