Research map / Large language models / Agents, code and math
Aria: an Agent for Retrieval and Iterative Auto-Formalization via Dependency Graph
Hanyu Wang, Ruohan Xie, Yutong Wang, Guoxiong Gao, XintaoYu, Bin Dong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add term-level definition grounding (AriaScorer vs LeanScorer) (ablation) | semantic checker · FATE-X semantic checking | accuracy: 71.0% -> 89.9% | Improved |
| add term-level definition grounding (AriaScorer vs LeanScorer) (ablation) | semantic checker · FATE-X semantic checking | recall: 88.5% -> 96.2% | Improved |
| raise AriaScorer threshold from 0 to 0.9 (ablation) | AriaScorer · FATE-X semantic checking | precision: 90.9% -> 95.5% | Improved |
| raise AriaScorer threshold from 0 to 0.9 (ablation) | AriaScorer · FATE-X semantic checking | recall: 96.2% -> 80.8% | Hurt |
| use Aria agent instead of Goedel-V2 at pass@128 (over 7x calls) | Aria · FATE-X | final accuracy: 24.0 -> 44.0 | Improved |
| increase Goedel-V2 samples from pass@16 to pass@128 | Goedel-V2 · FATE-X | compiler success rate: 37.0 -> 63.0 | Improved |
| add the Reflection module (ablation) | Aria · FATE-X and Conjectures | final accuracy | Improved |
| add the Graph-of-Thought planner (ablation) | Aria · Homological Conjectures | successful conjectures: 1 -> 6 | Improved |
| add the RAG retrieval module (ablation) | Aria · Homological Conjectures | final accuracy | Improved |
Closest papers by meaning
- APOLLO: Automated LLM and Lean Collaboration for Advanced Formal Reasoning (NeurIPS 2025 · similarity 0.88)
- HERMES: Towards Efficient and Verifiable Mathematical Reasoning in LLMs (ICML 2026 · similarity 0.87)
- APE-Bench: Evaluating Automated Proof Engineering for Formal Math Libraries (ICML 2026 · similarity 0.87)
- BAIT: Benchmarking (Embedding) Architectures for Interactive Theorem-Proving (AAAI 2024 · similarity 0.86)
- FaithSieve: Fine-Grained Evaluation of Math Proofs with Faithful Formal Evidence (NeurIPS 2026 · similarity 0.86)
- Magnushammer: A Transformer-Based Approach to Premise Selection (ICLR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
