Users praise Google DeepMind's Formal Conjectures Lean 4 Benchmark for AI Math because it facilitates new discoveries and bridges the formal math community with automated systems.
Based on 1 visible X reactions from 2 accounts; directional sample.
Ask a question below.
Published answers will appear here.
It’s wonderful to see that this work is already facilitating new discoveries and bridging the gap between the formal math community and automated solvers. Congratulations to everyone involved, especially former Pi colleague @MoritzFirsching, @Paul_lezeau, Salvatore Mercuri, and @mzhorvath for leading this work. Check out the open-source code here: https://github.com/google-deepmind/formal-conjectures
7:59 AM · Jul 14, 2026Thrilled to share our latest collaboration with @GoogleDeepMind: Formal Conjectures, an evolving Lean 4 benchmark for evaluating AI capabilities in formal math. https://arxiv.org/abs/2605.13171
7:59 AM · Jul 14, 2026Who wants to throw their AI agents at this one just for funsies https://github.com/google-deepmind/formal-conjectures/blob/main/FormalConjectures/Wikipedia/FibonacciPrimes.lean
7:59 PM · Jul 14, 2026I contributed one conjecture to this benchmark lol
Thrilled to share our latest collaboration with @GoogleDeepMind: Formal Conjectures, an evolving Lean 4 benchmark for evaluating AI capabilities in formal math. https://arxiv.org/abs/2605.13171
TBH I should check if it is solved yet, I genuinely wanted to know the answer (had to do with Fibbonaci primes)
7:55 PM · Jul 14, 2026I contributed one conjecture to this benchmark lol
Thrilled to share our latest collaboration with @GoogleDeepMind: Formal Conjectures, an evolving Lean 4 benchmark for evaluating AI capabilities in formal math. https://arxiv.org/abs/2605.13171
TBH I should check if it is solved yet, I genuinely wanted to know the answer (had to do with Fibbonaci primes)
7:55 PM · Jul 14, 2026Users praise Google DeepMind's Formal Conjectures Lean 4 Benchmark for AI Math because it facilitates new discoveries and bridges the formal math community with automated systems.
Based on 1 visible X reactions from 2 accounts; directional sample.
Ask a question below.
Published answers will appear here.