Demo code for "Revolutionizing Collection Service: How AI-Driven Personalization is Transforming Payment Recovery"

Overview

This repo contains demo code for the Medium post. It illustrates how one may make use to LLMs in collection services. The repo currently consists of four Jupyter notebooks, each showcasing one particular use case:

• plot_reasoning.ipynb: Plot-to-text generation and reasoning about the plot in plain text.
• credit_risk.ipynb: Credit risk assessment based on credit reports.
• rag.ipynb: Retrieval-augmented generation for LLMs to fetch the most relevant context before reasoning.
• guided_decoding.ipynb: Structured output generation through guided decoding given a template.

Note that some components of this code repo are still under development.

Installation

• Follow INSTALL.md to install the dependencies.

Contact

Xinchan Zhu ([email protected])