This repository includes:
- research and model fitting Jupyter Notebook pipeline, also available here
- Gradio web application
Final model configuration: MLP Classifier with 50 hidden layers and sigmoid activation function.
Trained with LBFGS optimizer, L2 regularization and early_stopping.
precision recall f1-score support
0 0.99 0.64 0.77 1495
1 0.04 0.71 0.08 34
accuracy 0.64 1529
macro avg 0.52 0.67 0.43 1529
weighted avg 0.97 0.64 0.76 1529
Accuracy Train Score: 0.6909937888198758
Accuracy Test Score: 0.6370176586003924
NDCG score: 0.43214574227842023
- Run
docker pull rakepants/assignment-app:latest
- Run
docker run -p 7860:7860 rakepants/assignment-app
- Navigate to http://localhost:7860/
- Clone this repository:
git clone https://github.com/RakePants/assignment
- Run
cd assignment
- Run
docker-compose up --build