-
Shi, Xiang authoredShi, Xiang authored
README.md 1.14 KiB
SWEMLS: Coursework 1: Training
This repository is mainly separated into 2 stages:
- Training:
train.py
- Inference:
model.py
They require different dependencies, so I separated them into 2 files.
requirements.txt is the dependencies for inferencing (i.e. running the model.py
)
requirements-dev.txt is the dependencies for training (i.e. running the train.py
)
Dependency | Environment | Description |
---|---|---|
torch | Inference & Training | This is for training and inferencing the ML model |
pandas | Inference & Training | This is for data preprocessing, this process and transform the csv file into the format that the model can use |
numpy | Inference & Training | Numerical computing library and help to do some data processing |
scikit-learn | Training | This provides some useful tool for training, e.g., train-test split and evaluation metrics |
early-stopping-pytorch | Training | This is for early stopping, it can help prevent overfitting |
Testing Result:
Metric | Value |
---|---|
Accuracy | 0.9976 |
Precision | 0.9974 |
Recall | 0.9910 |
F3 | 0.9917 |