Yinsheng Yao

Yinsheng Yao, Hongxiang Zhang, Weixi Tong et al.

Large language models often generate code with bugs. Existing methods rely on feedback signals such as test failures and self-critiques to iteratively refine the generated code. Such signals are either too coarse-grained or too high-level, which is not sufficient to inform the model where to fix the bug. In this work, we present Flare, an iterative framework with a lightweight diagnostic model that predicts line-level suspiciousness signals for bug localization and code refinement. Given the inherent uncertainty of diagnostic predictions, Flare searches over the top-k suspicious regions and selects the best candidate according to execution outcomes. Experiments on LiveCodeBench and BigCodeBench with five base LLMs show that, even without candidate search (k=1), Flare outperforms the strongest baseline with an absolute improvement from 1.72% to 7.42%. Furthermore, searching over 10 candidates yields an average improvement of 8.50% compared with no candidate search. When evaluated in isolation, our lightweight diagnostic model achieves the best performance compared with recent fault localization methods, demonstrating that it can provide reliable fine-grained guidance for code refinement.

Yiyun Zhu, Yidong Jiang, Ziwen Xu et al.

Large language models (LLMs) are increasingly used to generate financial research reports, shifting from auxiliary analytic tools to primary content producers. Yet recent real-world deployments reveal persistent failures--factual errors, numerical inconsistencies, fabricated references, and shallow analysis--that can distort assessments of corporate fundamentals and ultimately trigger severe economic losses. However, existing financial benchmarks focus on comprehension over completed reports rather than evaluating whether a model can produce reliable analysis. Moreover, current evaluation frameworks merely flag hallucinations and lack structured measures for deeper analytical skills, leaving key analytical bottlenecks undiscovered. To address these gaps, we introduce FinReasoning, a benchmark that decomposes Chinese research-report generation into three stages aligned with real analyst workflows, assessing semantic consistency, data alignment, and deep insight. We further propose a fine-grained evaluation framework that strengthens hallucination-correction assessment and incorporates a 12-indicator rubric for core analytical skills. Based on the evaluation results, FinReasoning reveals that most models exhibit a understanding-execution gap: they can identify errors but struggle to generate accurate corrections; they can retrieve data but have difficulty returning it in correct format. Furthermore, no model achieves overwhelming superiority across all three tracks; Doubao-Seed-1.8, GPT-5, and Kimi-K2 rank as the top three in overall performance, yet each exhibits a distinct capability distribution. The evaluation resource is available at https://github.com/TongjiFinLab/FinReasoning.

Yinsheng Yao, Jiehao Tang, Zhaozhen Yang et al.

While explicit reasoning trajectories enhance model interpretability, existing paradigms often rely on monolithic chains that lack intermediate verification, allowing early errors to cascade unchecked. This lack of modularity impedes granular auditing and compromises the epistemic trust required for high-stakes applications. We propose MAVEN (Multi-Agent Verification-Elaboration Network with In-Step Epistemic Auditing), a blackboard-inspired framework designed to transform LLMs into deliberate reasoners through explicit role-decoupling. At its core, MAVEN operationalizes an adversarial Skeptic-Researcher-Judge loop, simulating expert deliberation by functionally separating logical defense from factual grounding. Experiments on OpenBookQA, TruthfulQA, HALUEVAL and StrategyQA benchmarks demonstrate that MAVEN delivers superior reasoning quality across four fine-grained metrics. Notably, MAVEN consistently outperforms latent reasoning models such as GEMINI-3.1-Pro and consensus-based baselines (e.g., ReConcile) by generating explicitly structured, modular, and verifiable deliberation trajectories, rather than relying on implicit internal states or post-hoc consensus. Moreover, comprehensive evaluations confirm that MAVEN is fully model-agnostic, serving as a strong and transferable reasoning booster that yields substantial performance improvements across diverse backbone models.