Indu Indiramani

Title: Indu Indiramani and LLM Chunk-Size Optimization.

Introduction

Indu Indiramani is an inventor based in Bengaluru, India. Her patent work addresses how to optimize the size of data chunks processed by a large language model (LLM). She is associated with JPMorgan Chase Bank, N.A.

Latest Patents

Indu Indiramani is named on one patent, titled “Method and System for Optimizing a Chunk Size for a Large Language Model.” The invention describes receiving input data and evaluation metrics, dividing the data into chunks of predefined sizes, and processing each chunk through an LLM. It then calculates a comprehensive response quality index (CRQI) for each chunk as a weighted combination of evaluation metrics and selects the chunk with the highest CRQI as optimal. This approach provides a structured way to compare chunk sizes based on model output quality.

Career Highlights

The available information identifies Indu Indiramani as an inventor associated with JPMorgan Chase Bank, N.A. Her patent focuses on evaluating LLM performance through chunk-size comparisons and a combined quality measure.

Collaborations

Indu Indiramani’s listed coworker is Vipin Agarwal. The provided information does not specify further details about their collaboration.

Conclusion

Indu Indiramani’s patent presents a method for selecting data chunk sizes for large language models by comparing their outputs with a weighted quality index.