CrypFormBench is a new benchmark jointly covering symbolic and computational security to evaluate LLMs on five formal analysis capabilities, with results showing top model Claude-3.5 scores 48.7/100 and most models struggling on generation, transformation, and correction.
DT-Solver:AutomatedTheoremProvingwithDynamic-TreeSampling Guided by Proof-level Value Function
3 Pith papers cite this work, alongside 5 external citations. Polarity classification is still indexing.
3
Pith papers citing it
5
external citations · OpenAlex
years
2026 3verdicts
UNVERDICTED 3representative citing papers
Lean acts as a process-level reward oracle in GRPO-style RL, using tactic elaboration for first-error propagation and first-token credit to improve theorem proving over binary outcome rewards.
A minimal agentic system achieves competitive performance in automated theorem proving with a simpler design and lower cost than state-of-the-art methods.
citing papers explorer
-
CrypFormBench: Benchmarking Formal Analysis Capability of Large Language Models for Cryptographic Schemes
CrypFormBench is a new benchmark jointly covering symbolic and computational security to evaluate LLMs on five formal analysis capabilities, with results showing top model Claude-3.5 scores 48.7/100 and most models struggling on generation, transformation, and correction.
-
Process-Verified Reinforcement Learning for Theorem Proving via Lean
Lean acts as a process-level reward oracle in GRPO-style RL, using tactic elaboration for first-error propagation and first-token credit to improve theorem proving over binary outcome rewards.
-
A Minimal Agent for Automated Theorem Proving
A minimal agentic system achieves competitive performance in automated theorem proving with a simpler design and lower cost than state-of-the-art methods.