Aristotle achieves SOTA 96.8% proof generation on VERINA

Stream: Machine Learning for Theorem Proving

[Vikram Shanker]

We, at Harmonic, are happy to present our latest benchmark result! Evaluating Aristotle on the verification part of the VERINA benchmark, we have achieved 96.8%, proving 160 of the statements and disproving 23 of them by finding a counterexample. More details on this in our blog post at harmonic.fun/news.

[Notification Bot]

9 messages were moved from this topic to #Machine Learning for Theorem Proving > Aristotle writing rust? by Johan Commelin.