ConspiracyBench / inspect_eval /export_units.py

Commit History

28 days ago

  1. Baseline eval: five ~4B thinking models on ConspiracyBench, graded by pplx-decider (#428)⁠

    233d3a6