Tag
这篇论文研究了无标签评估方法（如自一致性、多数投票）中一致性与正确性之间的解耦现象，发现大多数错误答案集中在高一致性区域，从而对依赖一致性作为正确性代理的评估流程构成盲区。
FinSkillOps is a multi-agent system for SEC filing question answering that introduces controlled skill management for self-evolution, improving accuracy and reducing errors in financial QA systems.
CIFQA introduces a deterministic tool-grounded multi-agent LLM framework for financial query answering that separates linguistic interpretation from numerical execution, achieving high accuracy and outperforming larger models on calculation-intensive tasks.
This paper introduces HC-RAG, a hierarchical cross-modal retrieval-augmented generation framework for evidence-centric financial question answering over 10-K filings, along with a new benchmark Multi-Doc-2025. It outperforms RAPTOR and GraphRAG on financial QA benchmarks, especially for long-document and table-related queries.
This paper introduces FinAgent-RAG, an agentic framework for financial document question answering that combines iterative retrieval, Program-of-Thought reasoning, and adaptive resource allocation to improve accuracy and reduce costs.