MNIST Classification Model

A fully-connected neural network trained on the MNIST handwritten-digit dataset using PyTorch.
This notebook demonstrates data loading, model definition, training, evaluation, and visualization of results.

---

🚀 Features

• Data Loading via torchvision.datasets.MNIST
• Reproducible Training (fixed seeds for Python, NumPy, and PyTorch)
• Simple Feed-Forward Classifier using nn.Sequential and nn.Flatten
• Training & Validation Loop with loss tracking over epochs
• Evaluation Metrics:
  • Classification report (precision, recall, F1-score)
  • Confusion matrix visualized with Seaborn
• Device-Agnostic: runs on GPU if available, otherwise CPU