NLP and LLMs - Lab - Build a Question Answer (QA) Pipeline for Financial Documents
About
The goal of this lab is to build a Question Answer (QA) pipeline for financial documents that financial analysts can use to inform their research by processing tons of text based data quickly, and pulling out key insights.
- Note your results may be different from the solutions. Evaluate your results.
Steps:
- Build an initial pipeline using a single text document financial_context.
- Test a few pretrained models and select one to use for this task.
- Evaluate the performance of your model.
- Make recommendations for when if should and should not be used.
- BONUS: Extend the pipeline to a larger set of documents from the FinQA data.
Data
In this lab you will data from the FinQA dataset.
In the data folder you have a copy of the FinQA datasets.
Materials We Provide
| Topic | Description | Link |
|---|---|---|
| Activity | Build a Question Answer (QA) Pipeline for Financial Documents | Link |
Prerequisites
Before this activity, you should already be able to:
- Explore Hugging Face for NLP models
- Create an NLP pipeline in Hugging Face
Important note
Please ignore the _layouts folder and the config.yml file in the root of this repo. They are needed to make sure the repo renders well; please don’t change them or delete them and your repo will work fine :)