← back to claim
Held-out contamination audit found miniF2F-Lean4 gains persist at 54% on 300 freshly authored problems never posted online
cited by (1)
cited as supports by:
[+0]
Autoformalization success on undergraduate competition corpora rose from 8% to 61% between 2022 and 2025 on the miniF2F-Lean4 benchmark
centre graph here
[i]inspect this link
formal-verification
machine-learning
virtues: Accurate unrated · Falsifiable unrated · Clear unrated · Novel unrated · Important unrated
this claim:
Held-out contamination audit found miniF2F-Lean4 gains persist at 54% on 300 freshly authored problems never posted online
supports (1)
supports:
[+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
centre graph here
[i]inspect this link
formal-verification
reproducibility
virtues: Accurate unrated · Falsifiable unrated · Clear unrated · Novel unrated · Important unrated
opposes (0)
no opposing evidence yet
inspector
Select a link’s inspect marker to see who asserted it, why, and what it cites.