LESSON 0.7 · Assessment · 90 min
Mastery Gate 0: read a training loop
Mastery Gate 0: read a training loop. Learn the data, forward, loss, backward, and update loop and complete: Explain and repair three deliberately broken training loops. Part of th
Learning objectives
- Explain what problem “Mastery Gate 0: read a training loop” solves without hiding behind terminology.
- Trace the variables and causal links across the data, forward, loss.
- Complete “Explain and repair three deliberately broken training loops” and judge the result with evidence rather than intuition.
Core concepts
the data
the data is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment before moving to an optimized implementation.
forward
forward is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
loss
loss is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
backward
backward is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
and update loop
and update loop is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
Build and verify
Explain and repair three deliberately broken training loops
- Predict: write the expected output, trend, or failure before running code.
- Build: implement only the minimum components needed to answer the question.
- Verify: compare with a baseline or trusted implementation; save seeds, parameters, and raw outputs.
- Transfer: change one shape, dataset, scale, or workload condition and explain whether the conclusion still holds.