lifeomic-chatbot-tools 5.0.0

Utilities for machine learning, web services, and cloud infrastructure.

lifeomic-chatbot-tools

Python utilities for machine learning, web services, and cloud infrastructure. Includes classes and methods for:

1. ML model serialization/deserialization
2. ML model evaluation utilities
3. Data structures/models related to chatbots
4. ML model artifact persistence and version management
5. And more

The data structures in this package can all be found in the lifeomic_chatbot_tools.types sub-package, and are all Pydantic data models. For example the lifeomic_chatbot_tools.types.agent.AgentConfig class represents a chatbot's configuration and training data.

Getting Started

To begin using the package, use your favorite package manager to install it from PyPi. For example, using pip:

pip install lifeomic-chatbot-tools

Some of the features in this repo require more heavy weight dependencies, like AWS related utilities, or utilities specific to machine learning. If you try to import those features, they will tell you if you do not have the correct package extra installed. For example, many of the features in the lifeomic_chatbot_tools.ml sub-package require the ml extra. To install lifeomic-chatbot-tools with that extra:

pip install lifeomic-chatbot-tools[ml]

You can then begin using any package features that require ML dependencies.

Developing Locally

Before making any new commits or pull requests, please complete these steps.

1. Install the Poetry package manager for Python if you do not already have it. Installation instructions can be found here.
2. Clone the project.
3. From the root directory of the repo, install the dependencies, including all dev dependencies and extras:

   poetry install --extras "aws ml"
   

4. Install the pre-commit hooks, so they will run before each local commit. This includes linting, auto-formatting, and import sorting:

   pre-commit install
   

   Note: pre-commit is a dev dependency of the project, and is managed by Poetry. In order to use those dependencies from the terminal, you must be in a Poetry shell. This can be accomplished by first running poetry shell in your terminal.

Testing Locally

With Yarn, Docker, and docker-compose installed, run this command from the project root:

yarn ci-test

This will build the project, lint it, and run the unit tests and integration tests. All those steps can be run individually as well. See the scripts in the package.json file for the command names.

Releasing The Package

Releasing the package is automatically handled by CI, but three steps must be taken to trigger a successful release:

1. Use Poetry's version command to bump the package's version.
2. Commit and tag the repo with the exact same version the package was bumped to, e.g. v1.0.0 (include a preceding v).
3. Push the commit and tag to remote. These can be done together using: git push --atomic origin <branch_name> <tag>

CI will then build release the package to pypi with that version once the commit and tag are merged to the master branch.