Lina

@uchicago ‘27

[Haifeng Xu @haifengxu0]

Excited about CruxBench!

Folks might know our lab’s @ProphetArena which grades forecasting probabilities. CruxBench is our first effort to shift from grading probability numbers to grading reasoning INSIGHT!

Inspired by many conversations with domain experts — a constant theme I am hearing is that “I do not care about what numbers you give me and they can even be wrong, but show me your new insight that I can use to convince my VP” — that’s exactly what CRUX is for

With our amazing collaborators at @Research_FRI !

[Lina @linapiuc]

My first co-first paper as an undergraduate!! Excited to present at #NeurIPS2026!

[Amelia Dai @ameliadai_]

CruxBench is accepted to #NeurIPS2026! 🎉

We evaluate not the answers, but the questions worth asking: can LLMs discover the cruxes that move beliefs about the future?

🔗 https://t.co/sjruGnbSxe

with @linapiuc, @haifengxu0, and FRI team: @_elsehow, @nadja_flechner, @EzraKarger