Team: Magdalena Buszka, Bartosz Brzoza, Martyna Firgolska
git clone https://github.com/nerkulec/deepemotions.git
# git clone [email protected]:nerkulec/deepemotions.git
cd deepemotions
conda env create --file conda.yml
conda activate deepemotions
pip install -r requirements.txt
# Adding new package
pip install <package>
# pip install pipreqs
pipreqs .
kedro run
kedro run --pipeline=data_download
Assuming the data is already downloded
kedro run --pipeline=data_preprocess
kedro run --pipeline=model_download
Assuming the pre-trained model is already downloded and data is already downloaded and preprocessed
kedro run --pipeline=model_train
If not already logged before running the training log in to Wandb to save logs. You will need to input your key.
wandb login
The aim of this project is to classify emotions associated with comments. For training we used google's goemotions dataset.