TL;DR

COINS is an acronym coined by btatkerson to simplify how voice commands are understood. This repository shares the idea of a coin and stores the code for training the COINS NER model.

Training

Warning

Python 3.9 is the verified working version; any version, newer or older, may not function properly.

The COINS model is trained using spaCy; as such, this section is akin to their docs#usage. If you have trouble with spaCy, please review their docs before creating an issue here! To train the model, simply:

1. Install required dependencies

python3 -m pip install -r requirements.txt

2. Create a file for annotated data (more info at acoin)

echo "Please [set](COMMAND) a [timer](OPTION) for [15](NUMBER) [minutes](INPUT)" > annotations.txt

3. Run the training script

python3 main.py

This will export the model in the root directory under the name COINS, this can then be passed to Ellie for use. You may also want to test the model, which can be done by running

python3 src/test.py