Code used for "Training Agents to Self-Report Misbehavior"

A three-stage pipeline to collect realistic scheming behavior across varied domains, tool configurations, and task types.

bloom - evaluate any behavior immediately  🌸🌱

The premier resource for linguistics olympiads

A TeX tool for writing and formatting linguistics problems, based on International Linguistics Olympiad's format

语言学竞赛集成 / Collection on Linguistics Olympiad (Chinese version only)

A discord bot which can mock exams on discord. Currently mocking Physics Brawl 2019

A Discord bot based on discord.py which can conduct community problem solving marathons