← back to claim Autoformalization success on undergraduate competition corpora rose from 8% to 61% between 2022 and 2025 on the miniF2F-Lean4 benchmark

evidence graph

depth 1 2 3 both pro con
cited by (1)
[+0] By 2035 a machine-generated proof of a previously open Annals-level conjecture will be formally verified in Lean and accepted without a human-written proof sketch [i] ai-forecasting formal-verification mathematics
Autoformalization success on undergraduate competition corpora rose from 8% to 61% between 2022 and 2025 on the miniF2F-Lean4 benchmark
[+0] validity 0.0 centrality 17.6 consensus 0 depth 0.0
formal-verification machine-learning virtues
supports (2)
[+0] Proof length distribution on solved miniF2F problems is bounded above by 40 tactic steps, with a 96th-percentile of 22 [i] formal-verification
[+0] Held-out contamination audit found miniF2F-Lean4 gains persist at 54% on 300 freshly authored problems never posted online [i] formal-verification machine-learning
[+0] Independent re-evaluation by the Zurich formal methods group reproduced 51.8% on the held-out set with a clean-room rebuild of Kestrel-7 [i] formal-verification reproducibility
opposes (0)
no opposing evidence yet
whole evidence graph shown
inspector

Select a link’s marker to see who asserted it, why, and what it cites.