Skip to content

Fast Entry Point

Examples

Examples are the academy's 10 to 20 minute runs. Use them to confirm one idea, inspect one artifact, and decide whether you are ready for the full track. Do not try to run everything.

Pair With Topics Move To Tracks Decision Clinics Back To Home

Starter Pack

Learn The First Loop

Use a few small examples to make inspection, baselines, thresholds, and first decisions visible before opening a long workflow.

Comparison Pack

Change One Thing

The point of an example is not just to run. The point is to compare one change and see what evidence it created.

Pressure Pack

Pick The Next Move

Once the first runs are easy, use the mock-task and decision examples to practice restraint, review budgets, and stop rules.

Start With These Six

1. Shape Sanity

academy/.venv/bin/python academy/examples/numpy-shape-sanity/shape_sanity.py

Inspect:

  • which axis is being reduced
  • what shape each transformation produces
  • whether the stacked array still matches your expectation

Next: Array Shapes and Axis Operations

2. Course Support Inspection

academy/.venv/bin/python academy/examples/course-support-inspection/quick_inspection.py

Inspect:

  • missing values
  • suspicious columns
  • target balance
  • the first slice that looks risky

Next: Table Inspection

3. Honest Baseline Comparison

academy/.venv/bin/python academy/examples/validation-baseline-comparison/baseline_comparison.py

Inspect:

  • what the dummy baseline does
  • what the learned baseline actually improves
  • whether the split and preprocessing stayed honest

Next: Honest Splits and Baselines

4. Threshold Choices

academy/.venv/bin/python academy/examples/classical-ml-recipes/calibration_threshold_demo.py

Inspect:

  • ranking quality versus calibrated probability
  • which metric changes when the threshold moves
  • whether the threshold matches the real operating cost

Next: Calibration and Thresholds

5. First Training Loop

academy/.venv/bin/python academy/examples/deep-learning-recipes/pytorch_training_loop_demo.py

Inspect:

  • where training and evaluation are separated
  • what checkpoint would actually be saved
  • whether the learning curve tells you to continue or stop

Next: PyTorch Training Loops

6. Review Budget Decision

academy/.venv/bin/python academy/examples/mock-task-recipes/review_budget_demo.py

Inspect:

  • how coverage, precision, and review load trade off
  • whether the operating point is defensible
  • what queue policy you would actually choose

Next: Imbalanced Triage and Review Budgets

Pick An Example By Weakness

If data inspection is weak:

  • academy/examples/numpy-shape-sanity/shape_sanity.py
  • academy/examples/course-support-inspection/quick_inspection.py
  • academy/examples/pandas-slice-quicklook/slice_quicklook.py

If honest model comparison is weak:

  • academy/examples/validation-baseline-comparison/baseline_comparison.py
  • academy/examples/classical-ml-recipes/cross_validation_demo.py
  • academy/examples/classical-ml-recipes/leakage_patterns_demo.py

If thresholds and operating points are weak:

  • academy/examples/classical-ml-recipes/calibration_threshold_demo.py
  • academy/examples/mock-task-recipes/imbalance_metric_demo.py
  • academy/examples/mock-task-recipes/review_budget_demo.py

If deep-learning control is weak:

  • academy/examples/deep-learning-recipes/mlp_training_recipe.py
  • academy/examples/deep-learning-recipes/pytorch_training_loop_demo.py
  • academy/examples/deep-learning-recipes/optimizer_regularization_demo.py
  • academy/examples/deep-learning-recipes/transfer_finetuning_demo.py

If decision quality under pressure is weak:

  • academy/examples/mock-task-recipes/baseline_first_demo.py
  • academy/examples/mock-task-recipes/chronological_leaderboard_demo.py
  • academy/examples/decision-recipes/reliability_slices_demo.py
  • academy/examples/post-model-algorithm-recipes/ranking_assignment_demo.py

Route Maps

If you are on the beginner route:

  1. run academy/.venv/bin/python academy/examples/numpy-shape-sanity/shape_sanity.py
  2. run academy/.venv/bin/python academy/examples/course-support-inspection/quick_inspection.py
  3. move into Python, NumPy, Pandas, Visualization
  4. then run the matching lab and exercises

If you are on the IOAI route:

  1. run academy/.venv/bin/python academy/examples/validation-baseline-comparison/baseline_comparison.py
  2. run academy/.venv/bin/python academy/examples/classical-ml-recipes/calibration_threshold_demo.py
  3. do one clinic from Decision Clinics
  4. move into scikit-learn Validation and Tuning or Mock Tasks and Timed Workflows

Example Families

Use the family that matches the problem you are trying to repair.

  • Tooling and inspection: numpy-shape-sanity, course-support-inspection, course-support-baseline, pandas-slice-quicklook
  • Classical ML recipes: validation-baseline-comparison, classical-ml-recipes
  • Deep learning recipes: deep-learning-recipes
  • Decision and reliability recipes: decision-recipes, decision-clinic-recipes
  • Mock-task and competition recipes: mock-task-recipes
  • Advanced expansion: optimization-peft-recipes, representation-reuse-recipes, speech-audio-encoder-recipes, text-beyond-classification-recipes, unsupervised-manifold-recipes, problem-adaptation-recipes, post-model-algorithm-recipes, detection-segmentation-recipes

How To Use Examples Well

Use the same loop every time:

  1. predict what will matter in the output
  2. run one example
  3. inspect one table, plot, score, or checkpoint
  4. name the failure mode or decision point
  5. move to the matching topic or track

If you keep rerunning examples without writing down what changed, the page has stopped doing its job.