genai-pybcndata2023

Generative AI workshop delivered at PyDataBCN 2023

Role play

Help ResponsibleLending company to automate money loans to the right people, which are those aligning with ESG goals.
In order to do that we'll create a Gen-AI classification system to generate loan/no loan decisions according to two features:

1. email sentiment
2. ESG motivation of the sender

The system will also generate an automated email response for the sender extracting information such as:

1. sender
2. requested loan quantity
3. loan motivation

Seven loan request from different candidates are provided:

1. Lex Luthor
2. Thanos
3. The Joker
4. Agatha Christie
5. Aquaman
6. Coco Chanel
7. Salvador Dali

Content

This repo contains a notebook with code to:

1. Interact with models from HugginFace Inference API
2. Interact with a custom, fine-tuned ESG model from HuggingFace, ESG-BERT
3. Use prompts for Falcon-7b-instruct and flan-t5-xxl models models
4. Custom prompts to generate emails and their automated responses.