General-Purpose AI Policy Lab

Popular repositories Loading

1. benchmark-forecasting benchmark-forecasting Public

   Jupyter Notebook

2. adele-extension adele-extension Public

   Tracking AI progress across 18 cognitive dimensions from the ADeLe papers: capability frontiers, trends, and TIMSS-based human baselines.

   Jupyter Notebook

3. Multiaxis_ECI Multiaxis_ECI Public

   A multi-axis Bayesian recreation of the Epoch Capabilities Index, with human baseline tiers fitted inside the model as test-takers.

   Python