h2oGPTe GenAI Safety Events Classifier

This is an app which allows a user to upload a CSV file of safety event classifications generated using h2oGPTe GenAI LLM, provide human0in0the0loop feedback, and visually explore/manage the classification results and explanations.

End Users:

• Business Stakeholders - generate a classification using a GenAI LLM and explore the results and explanations.

Project Overview

├── data                    # Static files used primarily for local development and at run time
├── docs                    # Documentation
├── notebooks               # Jupyter notebooks for backend code development
├── src                     # Application source code
├── static                  # Non-changing assets
├── venv                    # Virtual python environment
├── .appignore              # Files to not publish to the AI Cloud
├── .gitignore              # Files to not check into GitHub
├── app.toml                # AI Cloud configuration
├── Makefile                # For local code compilation
├── README.md               
└── requirements.txt

Contributing

When working on this application, please complete the following items:

• Testing

  • Ensure that the app can run successfully locally
  • Ensure that the app can run in the AI Cloud using h2o bundle deploy

• Clean up

  • Add folders and files to the .gitignore which should not be checked into GitHub
  • Add folders and files to the .appignore which should do not need to be added to the AI Cloud
  • Run make format to ensure the codebase stays consistent using black and isort

• Documentation

  • Document any new App Secrets in the README.md
  • Add any new developer information to the README.md
  • Read the README.md, is everything still relevant and true?
  • Add any new user features to the docs/description_for_adverse_events_classifier.md
  • Read the docs/description_for_adverse_events_classifier.md, is everything still relevant and true?
  • Replace any dated images in the static folder

• Release

  • Increment the app version number in the app.toml
  • Create a new bundle using h2o bundle and this file into the build folder, no need to delete old bundles
  • Deploy the app to the AI Cloud

• Open a pull request (PR) in GitHub

  • Assign a reviewer
  • Assign at least one area label and type label
  • Use a descriptive title
  • Include a list of what your PR does in the description
  • Include RELNOTES=brief explaination in the description

Local Development

Setup your local python environment. Note that python 3.7 is specified as this is the default for the H2O AI Hybrid Cloud.

make setup

Run your app for local development after setting up the key information in Make File.

make run