The paper introduces a generalised prover-verifier game framework for 'neural interactive proofs', proves equilibria correspond to valid proof systems, and demonstrates modest gains in code validation with large language models.
The intuition behind the equilibrium (δp 3 , σv † ) is that the larger the probability mass on the setting when x = 3 (i.e
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Neural Interactive Proofs
The paper introduces a generalised prover-verifier game framework for 'neural interactive proofs', proves equilibria correspond to valid proof systems, and demonstrates modest gains in code validation with large language models.