cifar-10-alexnet

AlexNet on CIFAR-10 using PyTorch

This project implements AlexNet from scratch using PyTorch and trains it on the CIFAR-10 dataset.

The notebook covers:

• Loading and preprocessing the CIFAR-10 dataset
• Building AlexNet from scratch
• Training the model
• Evaluating test accuracy
• Saving the trained model
• Performing inference on sample images

Dataset

CIFAR-10 contains 60,000 color images of size 32×32 across 10 classes.

Classes:

• Airplane
• Automobile
• Bird
• Cat
• Deer
• Dog
• Frog
• Horse
• Ship
• Truck

Dataset path used:

/kaggle/input/datasets/bariankitvinod/cifar-10/data

Model

AlexNet (CIFAR-10 Version)

Architecture:

Conv → ReLU → MaxPool
Conv → ReLU → MaxPool
Conv → ReLU
Conv → ReLU
Conv → ReLU → MaxPool
Flatten
FC → ReLU → Dropout
FC → ReLU → Dropout
FC → Output (10)

Training

Loss Function

• CrossEntropyLoss

Optimizer

• SGD
• Momentum = 0.9
• Weight Decay = 5e-4

Learning Rate Scheduler

• StepLR

Results

Typical test accuracy:

Accuracy: 80.26%

Run

Open the notebook and execute all cells.

Project Structure

AlexNet/
│
├── alexnet.ipynb
├── README.md
├── requirements.txt
├── config.py
└── alexnet_cifar10.pth