We present FiMI (Finance Model for India), a domain-specialized financial language model developed for Indian digital payment systems. We develop two model variants: FiMI Base and FiMI Instruct. FiMI adapts the Mistral Small 24B architecture through a multi-stage training pipeline, beginning with continuous pre-training on 68 Billion tokens of curated financial, multilingual (English, Hindi, Hinglish), and synthetic data. This is followed by instruction fine-tuning and domain-specific supervised fine-tuning focused on multi-turn, tool-driven conversations that model real-world workflows, such as transaction disputes and mandate lifecycle management. Evaluations reveal that FiMI Base achieves a 20% improvement over the Mistral Small 24B Base model on finance reasoning benchmark, while FiMI Instruct outperforms the Mistral Small 24B Instruct model by 87% on domain-specific tool-calling. Moreover, FiMI achieves these significant domain gains while maintaining comparable performance to models of similar size on general benchmarks.

翻译：本文提出FiMI（面向印度的金融模型），这是一种为印度数字支付系统开发的领域专用金融语言模型。我们开发了两个模型变体：FiMI Base 和 FiMI Instruct。FiMI 基于 Mistral Small 24B 架构，通过多阶段训练流程进行适配。该流程始于对 680 亿个经过筛选的金融、多语言（英语、印地语、印英混合语）以及合成数据标记的持续预训练。随后进行指令微调和领域特定的监督微调，重点关注多轮次、工具驱动的对话，以模拟现实世界的工作流程，例如交易纠纷和授权生命周期管理。评估结果表明，在金融推理基准测试中，FiMI Base 相较于 Mistral Small 24B Base 模型实现了 20% 的性能提升；而在领域特定的工具调用任务上，FiMI Instruct 比 Mistral Small 24B Instruct 模型高出 87%。此外，FiMI 在取得这些显著领域性能提升的同时，在通用基准测试上仍保持了与同规模模型相当的性能。