Pinned Loading

1. prompt_rover prompt_rover Public

   Prompt Rover is an interactive tool for visualizing the latent space of an LLM. Given a prompt and its response, it builds a minimum path graph between embeddings and projects it with t-SNE to trac…

   Python 7 1

2. attribution-graph-probing attribution-graph-probing Public

   Automates attribution-graph analysis via probe prompting: circuit-trace a prompt, auto-generate concept probes, profile feature activations, cluster supernodes.

   Python 1

3. neuron-signatures neuron-signatures Public

   Neuron-signatures is a mechanistic interpretability pipeline for profiling MLP neurons via cross-prompt activation signatures and causal interventions. The method identifies stable functional roles…

   Python

4. ARENA_2.0 ARENA_2.0 Public

   Forked from callummcdougall/ARENA_2.0

   Resources for skilling up in AI alignment research engineering. Covers basics of deep learning, mechanistic interpretability, and RL.

   HTML

5. circuit-tracer circuit-tracer Public

   Forked from safety-research/circuit-tracer

   Python

6. TransformerLens TransformerLens Public

   Forked from TransformerLensOrg/TransformerLens

   A library for mechanistic interpretability of GPT-style language models

   Python