How ansshahzadd/neural-network-model-to-predict-handwritten-digits works
This is a small MNIST digit-classification training project, not a deployed inference service in the repository. Its main workflow loads MNIST, prepares image and label tensors, trains a CNN, evaluates test accuracy, and saves the model. The README also documents a Colab notebook route and links to a hosted drawing demo; the demo implementation is external. Internal stages share main.py because the script implements them together.
Main entry points: main.py Training script · train.ipynb Colab training
Components 11
Model training
- Training script
main.py - MNIST dataset
main.py - Data preparation
main.py - Image augmentation
main.py - CNN classifier
main.py - Model training
main.py - Test evaluation
main.py - Saved model
main.py - Colab training
train.ipynb
Hosted demo
- Hosted drawing demo
Other components
- Digit user
Connections 11
- Training script → MNIST dataset: loads
main.py - Training script → Data preparation: prepares
main.py - Training script → Image augmentation: configures
main.py - Training script → CNN classifier: defines
main.py - Training script → Model training: trains
main.py - Training script → Test evaluation: evaluates
main.py - Training script → Saved model: saves
main.py - Model training → CNN classifier: fits
main.py - Test evaluation → CNN classifier: tests
main.py - Digit user → Colab training: runs training
README.md - Digit user → Hosted drawing demo: draws digits
README.md