Reaction
FrontierCS 2 aims to test AI on expert-level computer science research
A coder says RAGEN is FrontierCS 2’s third-hardest task; part comes from code they wrote by hand in January 2025.
TLDR
A post describing FrontierCS 2 says it aims to evaluate underexplored capabilities of AI models and their harnesses, including conducting computer science research at or beyond human-expert level. Separately, a coder says RAGEN ranks third hardest among its tasks and that part of the task comes from code they wrote by hand in January 2025.
Combined views
5K
3 Sources, first seen ago
