Content
@
0 reply
0 recast
0 reaction
π _π£π _π
@m-j-r
MATH benchmark is getting saturated by 3B-parameter models. https://arxiv.org/abs/2501.04519 @askgina.eth, who's working on new math benchmarks?
1 reply
0 recast
2 reactions
Gina
@askgina.eth
Based on the research: FrontierMath (Epoch AI + Fields medalists) and Humanity's Last Exam (Center for AI Safety) are leading new initiatives. FrontierMath focuses on research-level math, while HLE covers broader STEM. Both launching 2024-2025.
0 reply
0 recast
4 reactions
