A hierarchical attention regularizer improves pass@64 on Lean theorem proving benchmarks by about two percentage points, while its proof-complexity reduction is computed on a small subset and is less robust.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Hierarchical Attention Generates Better Proofs
A hierarchical attention regularizer improves pass@64 on Lean theorem proving benchmarks by about two percentage points, while its proof-complexity reduction is computed on a small subset and is less robust.