PyTorch Neural Network from Scratch

Create a complete PyTorch neural network implementation for [TASK TYPE: classification/regression/etc.].

Requirements:
- Dataset: [DESCRIBE YOUR DATA]
- Input features: [NUMBER OF FEATURES]
- Output: [NUMBER OF CLASSES/CONTINUOUS VALUE]
- Preferred architecture: [MLP/CNN/RNN/Transformer]

Please provide:
1. Complete model class definition with configurable layers
2. Custom weight initialization
3. Forward pass implementation
4. Training loop with validation
5. Learning rate scheduling
6. Early stopping mechanism
7. Model checkpointing
8. Inference function
9. Performance visualization (loss curves, metrics)

Include proper device handling (CPU/GPU), gradient clipping, and batch processing.