FiMI: A Domain-Specific Language Model for Indian Finance Ecosystem

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.

Nowcasting the Financial Time Series with Streaming Data Analytics under Apache Spark

This paper proposes nowcasting of high-frequency financial datasets in real-time with a 5-minute interval using the streaming analytics feature of Apache Spark. The proposed 2 stage method consists of modelling chaos in the first stage and then using a sliding window approach for training with machine learning algorithms namely Lasso Regression, Ridge Regression, Generalised Linear Model, Gradient Boosting Tree and Random Forest available in the MLLib of Apache Spark in the second stage. For testing the effectiveness of the proposed methodology, 3 different datasets, of which two are stock markets namely National Stock Exchange & Bombay Stock Exchange, and finally One Bitcoin-INR conversion dataset. For evaluating the proposed methodology, we used metrics such as Symmetric Mean Absolute Percentage Error, Directional Symmetry, and Theil U Coefficient. We tested the significance of each pair of models using the Diebold Mariano (DM) test.