aijadugar/cifar-10-densenet

DenseNet on CIFAR-10

A PyTorch implementation of a DenseNet architecture trained from scratch on the CIFAR-10 dataset.

Model Details

• Architecture: DenseNet
• Framework: PyTorch
• Dataset: CIFAR-10
• Input Size: 3 × 32 × 32
• Classes: 10
• Growth Rate: 32

CIFAR-10 Classes

Label	Class
0	airplane
1	automobile
2	bird
3	cat
4	deer
5	dog
6	frog
7	horse
8	ship
9	truck

Training

• Optimizer: SGD
• Learning Rate: 0.1
• Momentum: 0.9
• Weight Decay: 5e-4
• Scheduler: StepLR
• Loss: CrossEntropyLoss
• Epochs: 30
• Batch Size: 128

Performance

Metric	Value
Test Accuracy	88.77%
Test Accuracy (DP)	88.77%

Model Files

• densenet_cifar10.pth

Load Model

model = DenseNet()

model.load_state_dict(
    torch.load("densenet_cifar10.pth")
)

model.eval()

Inference

with torch.no_grad():
    outputs = model(images)
    _, predicted = torch.max(outputs, 1)

Author

Ankit Bari

• GitHub: https://github.com/aijadugar
• Hugging Face: https://huggingface.co/aijadugar