peelml

peelml away the pain of ml deployment

Development

1. Setup and Build

1.1 build backend

git clone https://github.com/CambioML/peelml.git
conda create -n peelml python=3.10
conda activate peelml
cd peelml
pip install poetry
pip install git+https://github.com/UKPLab/sentence-transformers.git@179b659621c680371394d507683b25ba7faa0dd8
poetry install --no-root # no root to not install peelml as a package

1.2 build frontend

1. install nodejs
2. npm install npm install vite
3. cd into client folder and npm run build

2. Run

2.1 Option 1: run the server and client directly without cli

Before you start running the command below, go to server.py and upate

os.environ["OPENAI_API_KEY"] = ""

After that, run the following command

python -m peelml.server

2.2 Option 2: run the cli command

This is a interactive cli to ask you to input required environment parameters. It will then start the server and client for you.

python -m peelml.cli start

3. Install

install peelml into site-packages like a pip install

poetry install
peelml start

P.S: If you build and pip install for local test, your changes will not reflect in peelml site-packages. Therefore, you have to pip3 uninstall peelml, built it again, and the install.

Use

pip install git+https://github.com/UKPLab/sentence-transformers.git@179b659621c680371394d507683b25ba7faa0dd8
pip install peelml