When Financial Fine-tuning Fails: A Three-Level Detectability Analysis of Numeri

arXiv:2603.22305v1 Announce Type: cross Abstract: Large Language Models (LLMs) are rapidly transitioning from static Natural Language Processing (NLP) tasks including sentiment

arXiv:2605.04454v1 Announce Type: new Abstract: Alignment evaluation in machine learning has largely become evaluation of models. Influential benchmarks score model outputs under

arXiv:2606.17642v1 Announce Type: new Abstract: Financial multimodal reasoning requires agents to coordinate numerical computation, retrieval, visual interpretation, and temporal

arXiv:2609.00005v1 Announce Type: new Abstract: Financial scams targeting older adults increasingly occur through text and voice channels such as email, SMS, and phone calls,

arXiv:2605.04180v1 Announce Type: cross Abstract: Large Language Models exhibit strong reasoning and semantic understanding capabilities but often hallucinate in domains that

arXiv:2608.26114v1 Announce Type: new Abstract: Calculation-intensive financial question answering requires exact reasoning over structured rates, temporal conditions, numerical

arXiv:2610.24268v1 Announce Type: new Abstract: Form 10-Q MD&A sections combine long-form management narrative with accounting and forward-looking information. Benchmarks for