How kartikgupta98/dl-worksheets works
This repository is a deep-learning teaching site: learners open self-contained browser worksheets to explore concepts through explanations, exercises, quizzes, and interactive demonstrations. A separate fraud-competition branch supplies Kaggle learning materials and tooling: it generates synthetic transaction data, calibrates modeling configurations, exports competition files, and provides student and instructor notebooks and technique cards. The README documents the worksheet topics and competition handouts; source excerpts show the competition tooling’s main calls. Individual worksheet internals were not sampled, so their links describe the documented learner workflow rather than inspected code behavior.
Main entry points: index.html Worksheet index · calibrate.py Calibration runner · exp2.py Experiment configurations · pipeline.py Modeling pipeline
Components 23
Deep-learning worksheets
- Worksheet index
index.html - Activations
activations.html - Output-layer backprop
backprop_worksheet_output_layer.html - Gradient descent
lecture06_gradient_descent.html - Momentum and NAG
worksheet07_momentum_nag.html - Adaptive learning rates
worksheet08_adaptive_lr.html - Regularisation lab
worksheet09_regularisation.html - CNN worksheet
worksheet09_cnn.html - CNN architectures
worksheet10_architectures.html - Inception and ResNet
worksheet11_inception_resnet.html
Competition learning materials
- Technique cards
fraud_competition/materials/technique_cards.md - Student notebook
fraud_competition/materials/starter_notebook.ipynb - Instructor materials
fraud_competition/materials/instructor_guide.md
Data and calibration
- Synthetic data generator
fraud_competition/generate_data.py - Experiment configurations
fraud_competition/exp2.py - Modeling pipeline
fraud_competition/pipeline.py - Calibration runner
fraud_competition/calibrate.py - Experiment summary
fraud_competition/summary.py
Competition tooling
- Kaggle export and scoring
fraud_competition/export_kaggle.py - Notebook builder
fraud_competition/make_notebooks.py - Card-page renderer
fraud_competition/make_cards_html.py - Card-code notebook builder
fraud_competition/make_card_code_notebook.py
Other components
- Learner
Connections 23
- Learner → Worksheet index: opens
README.md - Worksheet index → Activations: links to
README.md - Worksheet index → Output-layer backprop: links to
README.md - Worksheet index → Gradient descent: links to
README.md - Worksheet index → Momentum and NAG: links to
README.md - Worksheet index → Adaptive learning rates: links to
README.md - Worksheet index → Regularisation lab: links to
README.md - Worksheet index → CNN worksheet: links to
README.md - Worksheet index → CNN architectures: links to
README.md - Worksheet index → Inception and ResNet: links to
README.md - Calibration runner → Synthetic data generator: generates data
fraud_competition/calibrate.py - Calibration runner → Experiment configurations: loads configs
fraud_competition/calibrate.py - Calibration runner → Modeling pipeline: runs models
fraud_competition/calibrate.py - Experiment configurations → Modeling pipeline: imports defaults
fraud_competition/exp2.py - Experiment summary → Calibration runner: uses summary
fraud_competition/summary.py - Kaggle export and scoring → Experiment configurations: uses settings
fraud_competition/export_kaggle.py - Kaggle export and scoring → Synthetic data generator: generates data
fraud_competition/export_kaggle.py - Kaggle export and scoring → Modeling pipeline: trains reference
fraud_competition/export_kaggle.py - Notebook builder → Modeling pipeline: uses pipeline
fraud_competition/make_notebooks.py - Notebook builder → Experiment configurations: uses configs
fraud_competition/make_notebooks.py - Card-page renderer → Technique cards: renders cards
fraud_competition/make_cards_html.py - Card-code notebook builder → Card-page renderer: uses card parser
fraud_competition/make_card_code_notebook.py - Card-code notebook builder → Notebook builder: uses notebook builder
fraud_competition/make_card_code_notebook.py