FinRAG-12B: A Production-Validated Recipe for Grounded Question Answering in Banking
For banking industry practitioners, this provides a production-validated recipe for deploying grounded, cost-effective LLMs with regulatory compliance.
The paper presents a data-efficient framework for training grounded LLMs in banking, achieving high answer quality with citation grounding outperforming GPT-4.1, a calibrated refusal mechanism reducing unsafe responses from 4.3% to 12%, and deployment at 40+ institutions with 7.1 percentage point improvement in query resolution and 3-5x faster responses at 20-50x lower cost.
Large language models (LLMs) are rapidly being adopted across various domains. However, their adoption in banking industry faces resistance due to demands for high accuracy, regulatory compliance, and the need for verifiable and grounded responses. We present a unified, data-efficient framework for training grounded domain-specific LLMs that optimizes answer quality, citation grounding, and calibrated refusal under real-world deployment constraints. First, we describe a data generation pipeline that combines LLM-as-a-Judge filtering, citation annotation, and curriculum learning with only 143M tokens. The resulting 12B model achieves high answer quality outperforming GPT-4.1 on citation grounding, with a modest citation tradeoff versus the untuned base. Second, we propose a calibrated refusal mechanism: training on 22% unanswerable examples yield a 12% "I don't know" rate, substantially improving over the base model's unsafe 4.3% rate while avoiding GPT-4.1's over-refusal (20.2%). Third, we present an end-to-end methodology spanning from data curation to quantized serving. The system is deployed at 40+ financial institutions, achieving a 7.1 percentage point improvement in query resolution (p < 0.001). Additionally, the model delivers 3-5x faster responses at 20-50x lower cost compared to GPT-4.1.