Eigen-1: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning
This work addresses inefficiencies in scientific reasoning for AI systems, offering a novel approach that improves accuracy and efficiency, though it is incremental in refining existing multi-agent and retrieval methods.
The paper tackled bottlenecks in scientific reasoning with LLMs, such as explicit retrieval disrupting reasoning and multi-agent pipelines diluting solutions, by introducing a unified framework with implicit retrieval and structured collaboration, achieving 48.3% accuracy on HLE Bio/Chem Gold, surpassing baselines by up to 18.1 points while reducing token usage by 53.5% and agent steps by 43.7%.
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.