About
I build Canonn, a context-faithful language model that answers strictly from the documents you give it - it cites the passage it used, and says "not in the provided data" instead of guessing. Most of my time goes into faithfulness evals and into making models that regulated teams can run on their own GPUs.
Badges
