PyTorch CNN implementation for MNIST digit classification.

Files

• MNIST_CNN_Classifier.ipynb - Main notebook with training and evaluation

Quick Start

1. Open in Google Colab
   • Note: Make a copy (File → Save a copy in Drive) before running
2. Run all cells

Results

• Test Accuracy: ~99%
• Model weights saved in saved_models/

/content/saved_models
├── accuracy_plot.png           # Training vs validation accuracy curves
├── loss_plot.png               # Training vs validation loss curves
├── classification_report.txt   # Class-wise precision, recall, F1-score
├── best_model.pth              # Checkpoint for resuming training
├── final_model_weights.pth     # Trained weights for inference/prediction
├── full_model.pth              # Complete model object
└── mnist_cnn_complete.pth      # Weights with metadata