VK ML Engineer Internship Assignment

This repository includes:

• research and model fitting Jupyter Notebook pipeline, also available here
• Gradio web application

Model

Final model configuration: MLP Classifier with 50 hidden layers and sigmoid activation function.
Trained with LBFGS optimizer, L2 regularization and early_stopping.

Metrics

              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

Launch

Variant 1

1. Run docker pull rakepants/assignment-app:latest
2. Run docker run -p 7860:7860 rakepants/assignment-app
3. Navigate to http://localhost:7860/

Variant 2

1. Clone this repository: git clone https://github.com/RakePants/assignment
2. Run cd assignment
3. Run docker-compose up --build