Saved in:
| Main Authors: | , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2509.21193 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912604939616256 |
|---|---|
| author | Tang, Xiangru Xu, Wanghan Wang, Yujie Guo, Zijie Shao, Daniel Chen, Jiapeng Zhang, Cixuan Wang, Ziyi Zhang, Lixin Wan, Guancheng Zhang, Wenlong Bai, Lei Yin, Zhenfei Torr, Philip Wang, Hanrui Jin, Di |
| author_facet | Tang, Xiangru Xu, Wanghan Wang, Yujie Guo, Zijie Shao, Daniel Chen, Jiapeng Zhang, Cixuan Wang, Ziyi Zhang, Lixin Wan, Guancheng Zhang, Wenlong Bai, Lei Yin, Zhenfei Torr, Philip Wang, Hanrui Jin, Di |
| contents | Large language models (LLMs) have recently shown strong progress on scientific reasoning, yet two major bottlenecks remain. First, explicit retrieval fragments reasoning, imposing a hidden "tool tax" of extra tokens and steps. Second, multi-agent pipelines often dilute strong solutions by averaging across all candidates. We address these challenges with a unified framework that combines implicit retrieval and structured collaboration. At its foundation, a Monitor-based retrieval module operates at the token level, integrating external knowledge with minimal disruption to reasoning. On top of this substrate, Hierarchical Solution Refinement (HSR) iteratively designates each candidate as an anchor to be repaired by its peers, while Quality-Aware Iterative Reasoning (QAIR) adapts refinement to solution quality. On Humanity's Last Exam (HLE) Bio/Chem Gold, our framework achieves 48.3\% accuracy -- the highest reported to date, surpassing the strongest agent baseline by 13.4 points and leading frontier LLMs by up to 18.1 points, while simultaneously reducing token usage by 53.5\% and agent steps by 43.7\%. Results on SuperGPQA and TRQA confirm robustness across domains. Error analysis shows that reasoning failures and knowledge gaps co-occur in over 85\% of cases, while diversity analysis reveals a clear dichotomy: retrieval tasks benefit from solution variety, whereas reasoning tasks favor consensus. Together, these findings demonstrate how implicit augmentation and structured refinement overcome the inefficiencies of explicit tool use and uniform aggregation. Code is available at: https://github.com/tangxiangru/Eigen-1. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_21193 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Eigen-1: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning Tang, Xiangru Xu, Wanghan Wang, Yujie Guo, Zijie Shao, Daniel Chen, Jiapeng Zhang, Cixuan Wang, Ziyi Zhang, Lixin Wan, Guancheng Zhang, Wenlong Bai, Lei Yin, Zhenfei Torr, Philip Wang, Hanrui Jin, Di Computation and Language Artificial Intelligence Large language models (LLMs) have recently shown strong progress on scientific reasoning, yet two major bottlenecks remain. First, explicit retrieval fragments reasoning, imposing a hidden "tool tax" of extra tokens and steps. Second, multi-agent pipelines often dilute strong solutions by averaging across all candidates. We address these challenges with a unified framework that combines implicit retrieval and structured collaboration. At its foundation, a Monitor-based retrieval module operates at the token level, integrating external knowledge with minimal disruption to reasoning. On top of this substrate, Hierarchical Solution Refinement (HSR) iteratively designates each candidate as an anchor to be repaired by its peers, while Quality-Aware Iterative Reasoning (QAIR) adapts refinement to solution quality. On Humanity's Last Exam (HLE) Bio/Chem Gold, our framework achieves 48.3\% accuracy -- the highest reported to date, surpassing the strongest agent baseline by 13.4 points and leading frontier LLMs by up to 18.1 points, while simultaneously reducing token usage by 53.5\% and agent steps by 43.7\%. Results on SuperGPQA and TRQA confirm robustness across domains. Error analysis shows that reasoning failures and knowledge gaps co-occur in over 85\% of cases, while diversity analysis reveals a clear dichotomy: retrieval tasks benefit from solution variety, whereas reasoning tasks favor consensus. Together, these findings demonstrate how implicit augmentation and structured refinement overcome the inefficiencies of explicit tool use and uniform aggregation. Code is available at: https://github.com/tangxiangru/Eigen-1. |
| title | Eigen-1: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2509.21193 |