Pytorch Learning

Simple Convolutional Neural Network (CNN) for CIFAR-10 image classification.

• Dataset: CIFAR-10 (10 classes of 32×32 RGB images)
• Model: Custom CNN with convolutional blocks, batch normalization, and dropout

---

🚀 Model Architecture

The network consists of:

1. Block 1

   • Conv2d(3→32, kernel=3, padding=1)
   • BatchNorm2d(32)
   • ReLU
   • MaxPool2d(2)

2. Block 2

   • Conv2d(32→32, kernel=3, padding=1)
   • BatchNorm2d(32)
   • ReLU
   • MaxPool2d(2)

3. Global Downsampling

   • Conv2d(32→64, kernel=3, padding=1)
   • BatchNorm2d(64)
   • ReLU
   • AdaptiveAvgPool2d(output_size=(2,2))

4. Deep Feature Block

   • Conv2d(64→128, kernel=3, padding=1)
   • Conv2d(128→256, kernel=3, padding=1)
   • BatchNorm2d(256)
   • ReLU

5. Classification Head

   • Flatten
   • Dropout(p=0.2)
   • Linear(256×2×2 → 256)
   • ReLU
   • Dropout(p=0.3)
   • Linear(256 → 64)
   • Linear(64 → 10)
   • Output logits for 10 CIFAR-10 classes

---

📈 Results

Accuracy: 86.41%