https://github.com/reneking-tech/cifar10-lightning

Hands-on deep learning with PyTorch & Lightning — CIFAR-10 classification
https://github.com/reneking-tech/cifar10-lightning

ai cifar10 cnn deep-learning image-classification machine-learning ml-engineering model-evaluation notebooks python pytorch pytorch-lightning transfer-learning

Last synced: about 2 months ago
JSON representation

Hands-on deep learning with PyTorch & Lightning — CIFAR-10 classification

• Host: GitHub
• URL: https://github.com/reneking-tech/cifar10-lightning
• Owner: reneking-tech
• Created: 2025-05-06T10:40:17.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2025-05-06T12:43:01.000Z (about 1 year ago)
• Last Synced: 2025-05-08T02:52:53.168Z (about 1 year ago)
• Topics: ai, cifar10, cnn, deep-learning, image-classification, machine-learning, ml-engineering, model-evaluation, notebooks, python, pytorch, pytorch-lightning, transfer-learning
• Language: Jupyter Notebook
• Homepage:
• Size: 112 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# 🧠 CIFAR-10 Classifier: From PyTorch to Lightning

This project demonstrates how to build and train a simple Convolutional Neural Network (CNN) on the CIFAR-10 image classification dataset using both **vanilla PyTorch** and **PyTorch Lightning**.

It was created as a practical, hands-on learning journey after completing Andrew Ng's Machine Learning Specialization. The goal: go from foundational ML understanding to real model implementation and training.

---

## 🚀 What’s Inside

- ✅ CIFAR-10 data loading with torchvision

- 🧱 CNN architecture (`SimpleCNN`)

- 🔁 Custom training + validation loops

- ⚡ PyTorch Lightning module (`LitCNN`)

- 📈 Loss/accuracy visualization

- 💡 Reflections on overfitting + improvements

- ✅ Pre-commit auto-formatting, notebook cleaning, code linting

---

## 📊 Training Results

### Vanilla PyTorch (5 Epochs)

- Final **Train Acc:** ~98%

- Final **Val Acc:** ~71%

- ❗ Validation loss increased → signs of **overfitting**

### PyTorch Lightning (5 Epochs)

- Final **Val Loss:** ~0.83

- Final **Val Accuracy:** ~71.2%

> 🔍 Same accuracy, but Lightning yielded a cleaner validation curve and easier setup.

---

## 🖼️ Training Curves

### 🔽 Loss

![Training vs Validation Loss](notebooks/images/loss_plot.png)

### 🔼 Accuracy

![Training vs Validation Accuracy](notebooks/images/accuracy_plot.png)

> _Outputs were stripped using pre-commit to keep this repo clean._

---

## 🧠 Reflections

Despite strong training performance, the validation metrics suggest overfitting. This provides a clear path to improvement:

- ✅ Add dropout layers

- ✅ Use weight decay

- ✅ Apply data augmentation (`RandomCrop`, `HorizontalFlip`)

- ✅ Explore more expressive architectures or learning rate schedulers

---

## ✅ Pre-Commit Enabled

![pre-commit](https://img.shields.io/badge/pre--commit-enabled-brightgreen?style=flat-square&logo=pre-commit)

This project uses [`pre-commit`](https://pre-commit.com/) to ensure:

- Clean `.ipynb` notebooks (`nbstripout`)

- Consistent code style (`black`, `isort`)

- Valid YAML + JSON

- No trailing whitespace or newline issues

To install:

```bash

pip install pre-commit

pre-commit install