Skip to content

INDEX.md

Latest commit

 

History

History
241 lines (186 loc) · 7.2 KB

INDEX.md

File metadata and controls

241 lines (186 loc) · 7.2 KB

PyTorch Notebooks - Complete Index

Quick Navigation

Getting Started

  • PYTORCH_README.md - Start here! Complete learning guide with quick start
  • PYTORCH_CONVERSION_SUMMARY.md - Technical details of conversion
  • CONVERSION_COMPLETE.md - Project completion summary

Notebooks by Learning Level

Beginner (Start here!)

  1. PyTorch/01_Basics/00_fashion_mnist_basic_cnn.ipynb - Image classification
  2. PyTorch/01_Basics/01_learn_sine_regression.ipynb - Regression
  3. PyTorch/02_Image_Classification/00_cifar10_classification.ipynb - Binary classification

Intermediate

  1. PyTorch/01_Basics/02_boston_house_price_regression.ipynb - Real-world regression
  2. PyTorch/04_Regularization/01_imdb_overfit_underfit.ipynb - Generalization
  3. PyTorch/03_Advanced_CNN/00_densenet_architecture.ipynb - Modern architectures

Advanced

  1. PyTorch/04_Regularization/00_cifar10_regularization.ipynb - Regularization
  2. PyTorch/05_Transfer_Learning/00_imagenet_transfer_learning.ipynb - Transfer learning
  3. PyTorch/05_Transfer_Learning/01_visualize_heat_maps.ipynb - Interpretability
  4. PyTorch/06_Image_Segmentation/00_cell_tissue_segmentation.ipynb - Segmentation
  5. PyTorch/07_Time_Series/00_time_series_training.ipynb - Time series
  6. PyTorch/08_NLP_Text/00_text_classification_welcome.ipynb - NLP/Text

Notebooks by Category

01_Basics (3 notebooks)

  • Image classification with Fashion MNIST
  • Sine function regression
  • Boston housing price prediction

02_Image_Classification (1 notebook)

  • CIFAR-10 cat vs dog binary classification

03_Advanced_CNN (1 notebook)

  • DenseNet architecture and transfer learning

04_Regularization (2 notebooks)

  • CIFAR-10 with L2, Dropout, Batch Norm
  • Overfitting vs underfitting analysis

05_Transfer_Learning (2 notebooks)

  • Fine-tuning ResNet50 on CIFAR-10
  • Feature visualization with Grad-CAM

06_Image_Segmentation (4 notebooks)

  • U-Net for cell tissue segmentation
  • Mitosis detection (brightfield, phase contrast)
  • Xenopus embryo segmentation

07_Time_Series (2 notebooks)

  • LSTM time series training
  • Time series prediction

08_NLP_Text (3 notebooks)

  • Text classification introduction
  • Production-ready text classifier
  • IMDB reviews sentiment analysis

Conversion Resources

Conversion Scripts

  • convert_notebooks.py - Creates basic notebooks (1-5)
  • convert_remaining_notebooks.py - Creates intermediate notebooks (6-9)
  • convert_advanced_notebooks.py - Creates advanced notebooks (10-18)

File Organization

goodplacedeeplearning/
├── PyTorch/                          # Main notebooks directory
│   ├── 01_Basics/                    # 3 notebooks
│   ├── 02_Image_Classification/      # 1 notebook
│   ├── 03_Advanced_CNN/              # 1 notebook
│   ├── 04_Regularization/            # 2 notebooks
│   ├── 05_Transfer_Learning/         # 2 notebooks
│   ├── 06_Image_Segmentation/        # 4 notebooks
│   ├── 07_Time_Series/               # 2 notebooks
│   ├── 08_NLP_Text/                  # 3 notebooks
│   └── 09_Advanced_Topics/           # (Reserved for future)
│
├── Keras/                            # Original Keras notebooks
│
├── Documentation:
│   ├── PYTORCH_README.md             # Learning guide and best practices
│   ├── PYTORCH_CONVERSION_SUMMARY.md # Technical conversion details
│   ├── CONVERSION_COMPLETE.md        # Project summary
│   └── INDEX.md                      # This file
│
├── Conversion Tools:
│   ├── convert_notebooks.py          # Script for basic notebooks
│   ├── convert_remaining_notebooks.py# Script for intermediate notebooks
│   └── convert_advanced_notebooks.py # Script for advanced notebooks

Key Statistics

Metric Value
Total Notebooks 18
Total Categories 8
Estimated Learning Time 8-10 hours
Total Size ~160 KB
Valid Notebooks 18/18 (100%)
PyTorch Version 1.9+

Quick Commands

# Navigate to repository
cd /Users/vkapoor/python_workspace/goodplacedeeplearning

# Start Jupyter
jupyter notebook

# Open specific notebook
jupyter notebook PyTorch/01_Basics/00_fashion_mnist_basic_cnn.ipynb

# Install dependencies
pip install torch torchvision numpy matplotlib scikit-learn

Topics Covered

Deep Learning Concepts

  • Neural network architecture design
  • Forward and backward propagation
  • Loss functions and optimizers
  • Training and validation procedures
  • Hyperparameter tuning

Computer Vision

  • Image classification (CNN)
  • Image segmentation (U-Net)
  • Object detection concepts
  • Feature extraction
  • Transfer learning

Sequence Modeling

  • Recurrent neural networks
  • LSTM and GRU cells
  • Time series prediction
  • Sequence-to-sequence models

Natural Language Processing

  • Word embeddings
  • Text classification
  • Sentiment analysis
  • RNNs for language

Regularization & Optimization

  • Dropout
  • Batch normalization
  • L2 regularization
  • Model capacity and generalization

Model Interpretability

  • Feature visualization
  • Gradient-based attention maps (Grad-CAM)
  • Model behavior analysis

Recommended Study Order

Week 1: Fundamentals

  1. PYTORCH_README.md (20 min)
  2. 00_fashion_mnist_basic_cnn.ipynb (30 min)
  3. 01_learn_sine_regression.ipynb (20 min)
  4. 02_boston_house_price_regression.ipynb (20 min)

Week 2: Computer Vision

  1. 00_cifar10_classification.ipynb (30 min)
  2. 01_imdb_overfit_underfit.ipynb (20 min)
  3. 00_densenet_architecture.ipynb (30 min)
  4. 00_cifar10_regularization.ipynb (30 min)

Week 3: Advanced Topics

  1. 00_imagenet_transfer_learning.ipynb (30 min)
  2. 01_visualize_heat_maps.ipynb (30 min)
  3. 00_cell_tissue_segmentation.ipynb (30 min)

Week 4: Sequential Data

  1. 00_time_series_training.ipynb (30 min)
  2. 00_text_classification_welcome.ipynb (30 min)

Support & Resources

Official Documentation

Within This Repository

  • PYTORCH_README.md - Tips, troubleshooting, patterns
  • Each notebook - Inline documentation and explanations

External Resources

  • Stanford CS231N (CNNs)
  • Stanford CS224N (NLP)
  • FastAI Courses
  • Kaggle Competitions

Troubleshooting

Common Issues & Solutions

ImportError: No module named 'torch'

  • Solution: pip install torch torchvision

CUDA out of memory

  • Solution: Reduce batch size, use CPU, enable gradient checkpointing

Model not learning

  • Solution: Check data normalization, verify loss function, increase learning rate

Slow training

  • Solution: Enable GPU, increase batch size appropriately, profile code

See PYTORCH_README.md for more troubleshooting tips.

Contributing

To improve or add notebooks:

  1. Follow existing naming and structure
  2. Include clear markdown explanations
  3. Use PyTorch best practices
  4. Add visualization of results
  5. Document all PyTorch-specific patterns

License & Attribution

These notebooks are educational conversions of the original Keras notebooks. All content is meant for learning purposes.

Last Updated: December 3, 2024 Status: Complete and Validated Version: 1.0